U.S. patent application number 17/509436 was filed with the patent office on 2022-05-19 for ghost point filtering.
The applicant listed for this patent is Motional AD LLC. Invention is credited to Thomas Koelbaek Jespersen, Yu Pan.
Application Number | 20220157015 17/509436 |
Document ID | / |
Family ID | 1000005925844 |
Filed Date | 2022-05-19 |
United States Patent
Application |
20220157015 |
Kind Code |
A1 |
Jespersen; Thomas Koelbaek ;
et al. |
May 19, 2022 |
GHOST POINT FILTERING
Abstract
Among other things, techniques are described for obtaining a
range image related to a depth sensor of a vehicle operating in an
environment. A first data point is identified in the range image
with an intensity at or below a first intensity threshold. A first
number of data points are determined in the range image that have
an intensity at or above a second intensity threshold in a first
region of the range image. Then, it is determined whether the first
number of data points is at or above a region number threshold. The
first data point is removed from the range image if the first
number of data points is at or above the region number threshold.
Operation of the vehicle is then facilitated in the environment
based at least in part on the range image. Other embodiments may be
described or claimed.
Inventors: |
Jespersen; Thomas Koelbaek;
(Singapore, SG) ; Pan; Yu; (Singapore,
SG) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Motional AD LLC |
Boston |
MA |
US |
|
|
Family ID: |
1000005925844 |
Appl. No.: |
17/509436 |
Filed: |
October 25, 2021 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16952046 |
Nov 18, 2020 |
11158120 |
|
|
17509436 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06T 17/00 20130101;
G06T 2207/10028 20130101; G01S 17/931 20200101; G06T 7/50
20170101 |
International
Class: |
G06T 17/00 20060101
G06T017/00; G01S 17/931 20060101 G01S017/931; G06T 7/50 20060101
G06T007/50 |
Claims
1. A method comprising: obtaining, using at least one processor, a
range image related to a depth sensor of a vehicle operating in an
environment; identifying, using the at least one processor, a first
data point in the range image with an intensity at or below a first
intensity threshold; determining, using the at least one processor,
a first number of data points in the range image that have an
intensity at or above a second intensity threshold in a first
region of the range image; determining, using the at least one
processor, whether the first number of data points is at or above a
region number threshold; removing, using the at least one
processor, the first data point from the range image if the first
number of data points is at or above the region number threshold;
and facilitating, using the at least one processor, operation of
the vehicle in the environment based at least in part on the range
image.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of U.S. application Ser.
No. 16/952,046, filed Nov. 18, 2020, now allowed, the entirety of
which is incorporated by reference.
FIELD OF THE INVENTION
[0002] This description relates to removal of ghost points from
point clouds produced by time-of-flight (ToF) systems such as a
light detection and ranging (LiDAR) system.
BACKGROUND
[0003] Typically, a ToF system such as LiDAR will collect ToF data
and generate a point cloud related to that data. The point cloud
may, in some cases, include points that are the result of
reflection errors within the transmitter or receiver of the ToF
system. In descriptions herein, these points are referred to as
"ghost" points, and their presence in the point cloud often causes
errors in the interpretation and processing of the ghost point and
the surrounding points of the point cloud. For example, an
autonomous vehicle that is identifying obstacles based on the point
cloud can mis-identify the presence of an object based on the ghost
points within the point cloud, which can negatively impact the
control of the vehicle.
BRIEF DESCRIPTION OF THE DRAWINGS
[0004] FIG. 1 shows an example of an autonomous vehicle having
autonomous capability.
[0005] FIG. 2 shows a computer system.
[0006] FIG. 3 shows an example architecture for an autonomous
vehicle.
[0007] FIG. 4 shows an example of inputs and outputs that can be
used by a perception system.
[0008] FIG. 5 shows an example of a LiDAR system.
[0009] FIG. 6 shows the LiDAR system in operation.
[0010] FIG. 7 shows the operation of the LiDAR system in additional
detail.
[0011] FIG. 8 shows an example comparison between a real-world
image and a point cloud which may result from a ToF system.
[0012] FIGS. 9a and 9b depict an example transmission structure and
receive structure of a ToF system.
[0013] FIG. 10 depicts an example of ghost point regions.
[0014] FIG. 11 depicts an example of a range image for processing
by a ghost point removal filter.
[0015] FIG. 12 depicts a high-level example technique of removal of
a ghost point from a range image.
[0016] FIG. 13 depicts a graphical example of the identification
and removal of a ghost point based on a ghost point region.
[0017] FIG. 14 depicts an example of a LiDAR pipeline.
[0018] FIG. 15 depicts an example of a technique related to
identifying and removing a ghost point from a point cloud.
DETAILED DESCRIPTION
[0019] In the following description, for the purposes of
explanation, numerous specific details are set forth in order to
provide a thorough understanding of the present disclosure. It will
be apparent, however, that the present disclosure may be practiced
without these specific details. In other instances, well-known
structures and devices are shown in block diagram form in order to
avoid unnecessarily obscuring the present disclosure.
[0020] In the drawings, specific arrangements or orderings of
schematic elements, such as those representing devices, systems,
instruction blocks and data elements, are shown for ease of
description. However, it should be understood by those skilled in
the art that the specific ordering or arrangement of the schematic
elements in the drawings is not meant to imply that a particular
order or sequence of processing, or separation of processes, is
required. Further, the inclusion of a schematic element in a
drawing is not meant to imply that such element is required in all
embodiments or that the features represented by such element may
not be included in or combined with other elements in some
embodiments.
[0021] Further, in the drawings, where connecting elements, such as
solid or dashed lines or arrows, are used to illustrate a
connection, relationship, or association between or among two or
more other schematic elements, the absence of any such connecting
elements is not meant to imply that no connection, relationship, or
association can exist. In other words, some connections,
relationships, or associations between elements are not shown in
the drawings so as not to obscure the disclosure. In addition, for
ease of illustration, a single connecting element is used to
represent multiple connections, relationships or associations
between elements. For example, where a connecting element
represents a communication of signals, data, or instructions, it
should be understood by those skilled in the art that such element
represents one or multiple signal paths (e.g., a bus), as may be
needed, to affect the communication.
[0022] Reference will now be made in detail to embodiments,
examples of which are illustrated in the accompanying drawings. In
the following detailed description, numerous specific details are
set forth in order to provide a thorough understanding of the
various described embodiments. However, it will be apparent to one
of ordinary skill in the art that the various described embodiments
may be practiced without these specific details. In other
instances, well-known methods, procedures, components, circuits,
and networks have not been described in detail so as not to
unnecessarily obscure aspects of the embodiments.
[0023] Several features are described hereafter that can each be
used independently of one another or with any combination of other
features. However, any individual feature may not address any of
the problems discussed above or might only address one of the
problems discussed above. Some of the problems discussed above
might not be fully addressed by any of the features described
herein. Although headings are provided, information related to a
particular heading, but not found in the section having that
heading, may also be found elsewhere in this description.
Embodiments are described herein according to the following
outline:
[0024] 1. General Overview
[0025] 2. System Overview
[0026] 3. Autonomous Vehicle Architecture
[0027] 4. Autonomous Vehicle Inputs
[0028] 5. Ghost Point Occurrence
[0029] 6. Ghost Point Filter Techniques
General Overview
[0030] A vehicle (such as an autonomous vehicle) will include at
least one processor that uses a ghost point filter to remove ghost
points from a three-dimensional (3D) point cloud generated by a
LiDAR system or some other ToF system (e.g., radio detection and
ranging (RADAR) or some other system) included in or on the
vehicle. As noted above, the ghost points are erroneous points of a
point cloud captured by the LiDAR system during a LiDAR scan that
do not correspond to an actual object reflection but rather are the
result of an optical artifact. In an embodiment, the ghost point
filter identifies a low-intensity point (e.g., a point with a low
optical intensity, which may also be considered as an optical
return with low received energy) in a point cloud. The ghost point
filter identifies the number of high-intensity points (e.g., a
point with a high optical intensity, which may also be considered
as an optical return that saturates the receiver) in pre-defined
regions of the point cloud that are relative to the identified
low-intensity point. Based on the number of high-intensity points
in the pre-defined regions, the low-intensity point may be
identified as a ghost point, and removed from the point cloud, or
kept in the point cloud. It will be understood that embodiments
herein will be described with respect to LiDAR, however another
embodiment will additionally or alternatively include some other
type of ToF detection.
[0031] An advantage of removing ghost points using the disclosed
ghost point filter is that the accuracy of the point cloud, and
resultant operation of the vehicle which relies in part on an
accurate point cloud, is improved. Notably, this advantage is
realized through a software solution that may otherwise require
expensive hardware modification to a significant number of LiDAR
systems. In an embodiment, the software solution will be recognized
as computationally efficient, as it is designed to operate on an
initial range image of the system (e.g., a range image prior to any
post-processing of the range image produced by the LiDAR
system(s)).
System Overview
[0032] FIG. 1 shows an example of an autonomous vehicle (AV) 100
having autonomous capability.
[0033] As used herein, the term "autonomous capability" refers to a
function, feature, or facility that enables a vehicle to be
partially or fully operated without real-time human intervention,
including without limitation fully autonomous vehicles, highly
autonomous vehicles, and conditionally autonomous vehicles.
[0034] As used herein, an AV is a vehicle that possesses autonomous
capability.
[0035] As used herein, "vehicle" includes means of transportation
of goods or people. For example, cars, buses, trains, airplanes,
drones, trucks, boats, ships, submersibles, dirigibles, etc. A
driverless car is an example of a vehicle.
[0036] As used herein, "trajectory" refers to a path or route to
navigate an AV from a first spatiotemporal location to second
spatiotemporal location. In an embodiment, the first spatiotemporal
location is referred to as the initial or starting location and the
second spatiotemporal location is referred to as the destination,
final location, goal, goal position, or goal location. In some
examples, a trajectory is made up of one or more segments (e.g.,
sections of road) and each segment is made up of one or more blocks
(e.g., portions of a lane or intersection). In an embodiment, the
spatiotemporal locations correspond to real-world locations. For
example, the spatiotemporal locations are pick up or drop-off
locations to pick up or drop-off persons or goods.
[0037] As used herein, "sensor(s)" includes one or more hardware
components that detect information about the environment
surrounding the sensor. Some of the hardware components can include
sensing components (e.g., image sensors, biometric sensors),
transmitting and/or receiving components (e.g., laser or radio
frequency wave transmitters and receivers), electronic components
such as analog-to-digital converters, a data storage device (such
as a random-access memory (RAM) and/or a non-volatile storage),
software or firmware components and data processing components such
as an ASIC (application-specific integrated circuit), a
microprocessor and/or a microcontroller.
[0038] As used herein, a "scene description" is a data structure
(e.g., list) or data stream that includes one or more classified or
labeled objects detected by one or more sensors on the AV vehicle
or provided by a source external to the AV.
[0039] As used herein, a "road" is a physical area that can be
traversed by a vehicle, and may correspond to a named thoroughfare
(e.g., city street, interstate freeway, etc.) or may correspond to
an unnamed thoroughfare (e.g., a driveway in a house or office
building, a section of a parking lot, a section of a vacant lot, a
dirt path in a rural area, etc.). Because some vehicles (e.g.,
4-wheel-drive pickup trucks, sport utility vehicles, etc.) are
capable of traversing a variety of physical areas not specifically
adapted for vehicle travel, a "road" may be a physical area not
formally defined as a thoroughfare by any municipality or other
governmental or administrative body.
[0040] As used herein, a "lane" is a portion of a road that can be
traversed by a vehicle. A lane is sometimes identified based on
lane markings. For example, a lane may correspond to most or all of
the space between lane markings, or may correspond to only some
(e.g., less than 50%) of the space between lane markings. For
example, a road having lane markings spaced far apart might
accommodate two or more vehicles between the markings, such that
one vehicle can pass the other without traversing the lane
markings, and thus could be interpreted as having a lane narrower
than the space between the lane markings, or having two lanes
between the lane markings. A lane could also be interpreted in the
absence of lane markings. For example, a lane may be defined based
on physical features of an environment, e.g., rocks and trees along
a thoroughfare in a rural area or, e.g., natural obstructions to be
avoided in an undeveloped area. A lane could also be interpreted
independent of lane markings or physical features. For example, a
lane could be interpreted based on an arbitrary path free of
obstructions in an area that otherwise lacks features that would be
interpreted as lane boundaries. In an example scenario, an AV could
interpret a lane through an obstruction-free portion of a field or
empty lot. In another example scenario, an AV could interpret a
lane through a wide (e.g., wide enough for two or more lanes) road
that does not have lane markings. In this scenario, the AV could
communicate information about the lane to other AVs so that the
other AVs can use the same lane information to coordinate path
planning among themselves.
[0041] "One or more" includes a function being performed by one
element, a function being performed by more than one element, e.g.,
in a distributed fashion, several functions being performed by one
element, several functions being performed by several elements, or
any combination of the above.
[0042] It will also be understood that, although the terms first,
second, etc. are, in some instances, used herein to describe
various elements, these elements should not be limited by these
terms. These terms are only used to distinguish one element from
another. For example, a first contact could be termed a second
contact, and, similarly, a second contact could be termed a first
contact, without departing from the scope of the various described
embodiments. The first contact and the second contact are both
contacts, but they are not the same contact.
[0043] The terminology used in the description of the various
described embodiments herein is for the purpose of describing
particular embodiments only and is not intended to be limiting. As
used in the description of the various described embodiments and
the appended claims, the singular forms "a," "an" and "the" are
intended to include the plural forms as well, unless the context
clearly indicates otherwise. It will also be understood that the
term "and/or" as used herein refers to and encompasses any and all
possible combinations of one or more of the associated listed
items. It will be further understood that the terms "includes,"
"including," "comprises," and/or "comprising," when used in this
description, specify the presence of stated features, integers,
steps, operations, elements, and/or components, but do not preclude
the presence or addition of one or more other features, integers,
steps, operations, elements, components, and/or groups thereof.
[0044] As used herein, the term "if" is, optionally, construed to
mean "when" or "upon" or "in response to determining" or "in
response to detecting," depending on the context. Similarly, the
phrase "if it is determined" or "if [a stated condition or event]
is detected" is, optionally, construed to mean "upon determining"
or "in response to determining" or "upon detecting [the stated
condition or event]" or "in response to detecting [the stated
condition or event]," depending on the context.
[0045] As used herein, an AV system refers to the AV along with the
array of hardware, software, stored data, and data generated in
real-time that supports the operation of the AV. In an embodiment,
the AV system is incorporated within the AV. In an embodiment, the
AV system is spread across several locations. For example, some of
the software of the AV system is implemented on a cloud computing
environment.
[0046] In general, this document describes technologies applicable
to any vehicles that have one or more autonomous capabilities
including fully AVs, highly AVs, and conditionally autonomous
vehicles, such as so-called Level 5, Level 4 and Level 3 vehicles,
respectively (see SAE International's standard J3016: Taxonomy and
Definitions for Terms Related to On-Road Motor Vehicle Automated
Driving Systems, which is incorporated by reference in its
entirety, for more details on the classification of levels of
autonomy in vehicles). The technologies described in this document
are also applicable to partially autonomous vehicles and driver
assisted vehicles, such as so-called Level 2 and Level 1 vehicles
(see SAE International's standard J3016: Taxonomy and Definitions
for Terms Related to On-Road Motor Vehicle Automated Driving
Systems). In an embodiment, one or more of the Level 1, 2, 3, 4 and
5 vehicle systems may automate certain vehicle operations (e.g.,
steering, braking, and using maps) under certain operating
conditions based on processing of sensor inputs. The technologies
described in this document can benefit vehicles in any levels,
ranging from fully autonomous vehicles to human-operated
vehicles.
[0047] Autonomous vehicles have advantages over vehicles that
require a human driver. One advantage is safety. For example, in
2016, the United States experienced 6 million automobile accidents,
2.4 million injuries, 40,000 fatalities, and 13 million vehicles in
crashes, estimated at a societal cost of $910+ billion. U.S.
traffic fatalities per 100 million miles traveled have been reduced
from about six to about one from 1965 to 2015, in part due to
additional safety measures deployed in vehicles. For example, an
additional half second of warning that a crash is about to occur is
believed to mitigate 60% of front-to-rear crashes. However, passive
safety features (e.g., seat belts, airbags) have likely reached
their limit in improving this number. Thus, active safety measures,
such as automated control of a vehicle, are the likely next step in
improving these statistics. Because human drivers are believed to
be responsible for a critical pre-crash event in 95% of crashes,
automated driving systems are likely to achieve better safety
outcomes, e.g., by reliably recognizing and avoiding critical
situations better than humans; making better decisions, obeying
traffic laws, and predicting future events better than humans; and
reliably controlling a vehicle better than a human.
[0048] Referring to FIG. 1, an AV system 120 operates the AV 100
along a trajectory 198 through an environment 190 to a destination
199 (sometimes referred to as a final location) while avoiding
objects (e.g., natural obstructions 191, vehicles 193, pedestrians
192, cyclists, and other obstacles) and obeying rules of the road
(e.g., rules of operation or driving preferences).
[0049] In an embodiment, the AV system 120 includes devices 101
that are instrumented to receive and act on operational commands
from the computer processors 146. We use the term "operational
command" to mean an executable instruction (or set of instructions)
that causes a vehicle to perform an action (e.g., a driving
maneuver). Operational commands can, without limitation, including
instructions for a vehicle to start moving forward, stop moving
forward, start moving backward, stop moving backward, accelerate,
decelerate, perform a left turn, and perform a right turn. In an
embodiment, computing processors 146 are similar to the processor
204 described below in reference to FIG. 2. Examples of devices 101
include a steering control 102, brakes 103, gears, accelerator
pedal or other acceleration control mechanisms, windshield wipers,
side-door locks, window controls, and turn-indicators.
[0050] In an embodiment, the AV system 120 includes sensors 121 for
measuring or inferring properties of state or condition of the AV
100, such as the AV's position, linear and angular velocity and
acceleration, and heading (e.g., an orientation of the leading end
of AV 100). Example of sensors 121 are GPS, inertial measurement
units (IMU) that measure both vehicle linear accelerations and
angular rates, wheel speed sensors for measuring or estimating
wheel slip ratios, wheel brake pressure or braking torque sensors,
engine torque or wheel torque sensors, and steering angle and
angular rate sensors.
[0051] In an embodiment, the sensors 121 also include sensors for
sensing or measuring properties of the AV's environment. For
example, monocular or stereo video cameras 122 in the visible
light, infrared or thermal (or both) spectra, LiDAR 123, RADAR,
ultrasonic sensors, ToF depth sensors, speed sensors, temperature
sensors, humidity sensors, and precipitation sensors.
[0052] In an embodiment, the AV system 120 includes a data storage
unit 142 and memory 144 for storing machine instructions associated
with computer processors 146 or data collected by sensors 121. In
an embodiment, the data storage unit 142 is similar to the ROM 208
or storage device 210 described below in relation to FIG. 2. In an
embodiment, memory 144 is similar to the main memory 206 described
below. In an embodiment, the data storage unit 142 and memory 144
store historical, real-time, and/or predictive information about
the environment 190. In an embodiment, the stored information
includes maps, driving performance, traffic congestion updates or
weather conditions. In an embodiment, data relating to the
environment 190 is transmitted to the AV 100 via a communications
channel from a remotely located database 134.
[0053] In an embodiment, the AV system 120 includes communications
devices 140 for communicating measured or inferred properties of
other vehicles' states and conditions, such as positions, linear
and angular velocities, linear and angular accelerations, and
linear and angular headings to the AV 100. These devices include
Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I)
communication devices and devices for wireless communications over
point-to-point or ad hoc networks or both. In an embodiment, the
communications devices 140 communicate across the electromagnetic
spectrum (including radio and optical communications) or other
media (e.g., air and acoustic media). A combination of
Vehicle-to-Vehicle (V2V) Vehicle-to-Infrastructure (V2I)
communication (and, in some embodiments, one or more other types of
communication) is sometimes referred to as Vehicle-to-Everything
(V2X) communication. V2X communication typically conforms to one or
more communications standards for communication with, between, and
among autonomous vehicles.
[0054] In an embodiment, the communication devices 140 include
communication interfaces. For example, wired, wireless, WiMAX,
Wi-Fi, Bluetooth, satellite, cellular, optical, near field,
infrared, or radio interfaces. The communication interfaces
transmit data from a remotely located database 134 to AV system
120. In an embodiment, the remotely located database 134 is
embedded in a cloud computing environment. The communication
devices 140 transmit data collected from sensors 121 or other data
related to the operation of AV 100 to the remotely located database
134. In an embodiment, communication devices 140 transmit
information that relates to teleoperations to the AV 100. In some
embodiments, the AV 100 communicates with other remote (e.g.,
"cloud") servers 136.
[0055] In an embodiment, the remotely located database 134 also
stores and transmits digital data (e.g., storing data such as road
and street locations). Such data is stored on the memory 144 on the
AV 100, or transmitted to the AV 100 via a communications channel
from the remotely located database 134.
[0056] In an embodiment, the remotely located database 134 stores
and transmits historical information about driving properties
(e.g., speed and acceleration profiles) of vehicles that have
previously traveled along trajectory 198 at similar times of day.
In one implementation, such data can be stored on the memory 144 on
the AV 100, or transmitted to the AV 100 via a communications
channel from the remotely located database 134.
[0057] Computer processors 146 located on the AV 100
algorithmically generate control actions based on both real-time
sensor data and prior information, allowing the AV system 120 to
execute its autonomous driving capabilities.
[0058] In an embodiment, the AV system 120 includes computer
peripherals 132 coupled to computer processors 146 for providing
information and alerts to, and receiving input from, a user (e.g.,
an occupant or a remote user) of the AV 100. In an embodiment,
peripherals 132 are similar to the display 212, input device 214,
and cursor controller 216 discussed below in reference to FIG. 2.
The coupling is wireless or wired. Any two or more of the interface
devices can be integrated into a single device.
[0059] In an embodiment, the AV system 120 receives and enforces a
privacy level of a passenger, e.g., specified by the passenger or
stored in a profile associated with the passenger. The privacy
level of the passenger determines how particular information
associated with the passenger (e.g., passenger comfort data,
biometric data, etc.) is permitted to be used, stored in the
passenger profile, and/or stored on the cloud server 136 and
associated with the passenger profile. In an embodiment, the
privacy level specifies particular information associated with a
passenger that is deleted once the ride is completed. In an
embodiment, the privacy level specifies particular information
associated with a passenger and identifies one or more entities
that are authorized to access the information. Examples of
specified entities that are authorized to access information can
include other AVs, third-party AV systems, or any entity that could
potentially access the information.
[0060] A privacy level of a passenger can be specified at one or
more levels of granularity. In an embodiment, a privacy level
identifies specific information to be stored or shared. In an
embodiment, the privacy level applies to all the information
associated with the passenger such that the passenger can specify
that none of her personal information is stored or shared.
Specification of the entities that are permitted to access
particular information can also be specified at various levels of
granularity. Various sets of entities that are permitted to access
particular information can include, for example, other AVs, cloud
servers 136, specific third-party AV systems, etc.
[0061] In an embodiment, the AV system 120 or the cloud server 136
determines if certain information associated with a passenger can
be accessed by the AV 100 or another entity. For example, a
third-party AV system that attempts to access passenger input
related to a particular spatiotemporal location must obtain
authorization, e.g., from the AV system 120 or the cloud server
136, to access the information associated with the passenger. For
example, the AV system 120 uses the passenger's specified privacy
level to determine whether the passenger input related to the
spatiotemporal location can be presented to the third-party AV
system, the AV 100, or to another AV. This enables the passenger's
privacy level to specify which other entities are allowed to
receive data about the passenger's actions or other data associated
with the passenger.
[0062] FIG. 2 shows a computer system 200. In an implementation,
the computer system 200 is a special-purpose computing device. The
special-purpose computing device is hard-wired to perform the
techniques or includes digital electronic devices such as one or
more ASICs or field programmable gate arrays (FPGAs) that are
persistently programmed to perform the techniques, or can include
one or more general-purpose hardware processors programmed to
perform the techniques pursuant to program instructions in
firmware, memory, other storage, or a combination. Such
special-purpose computing devices can also combine custom
hard-wired logic, ASICs, or FPGAs with custom programming to
accomplish the techniques. In various embodiments, the
special-purpose computing devices are desktop computer systems,
portable computer systems, handheld devices, network devices or any
other device that incorporates hard-wired and/or program logic to
implement the techniques.
[0063] In an embodiment, the computer system 200 includes a bus 202
or other communication mechanism for communicating information, and
a processor 204 coupled with a bus 202 for processing information.
The processor 204 is, for example, a general-purpose
microprocessor. The computer system 200 also includes a main memory
206, such as a RAM or other dynamic storage device, coupled to the
bus 202 for storing information and instructions to be executed by
processor 204. In one implementation, the main memory 206 is used
for storing temporary variables or other intermediate information
during execution of instructions to be executed by the processor
204. Such instructions, when stored in non-transitory storage media
accessible to the processor 204, render the computer system 200
into a special-purpose machine that is customized to perform the
operations specified in the instructions.
[0064] In an embodiment, the computer system 200 further includes a
read only memory (ROM) 208 or other static storage device coupled
to the bus 202 for storing static information and instructions for
the processor 204. A storage device 210, such as a magnetic disk,
optical disk, solid-state drive, or three-dimensional cross point
memory is provided and coupled to the bus 202 for storing
information and instructions.
[0065] In an embodiment, the computer system 200 is coupled via the
bus 202 to a display 212, such as a cathode ray tube (CRT), a
liquid crystal display (LCD), plasma display, light emitting diode
(LED) display, or an organic light emitting diode (OLED) display
for displaying information to a computer user. An input device 214,
including alphanumeric and other keys, is coupled to bus 202 for
communicating information and command selections to the processor
204. Another type of user input device is a cursor controller 216,
such as a mouse, a trackball, a touch-enabled display, or cursor
direction keys for communicating direction information and command
selections to the processor 204 and for controlling cursor movement
on the display 212. This input device typically has two degrees of
freedom in two axes, a first axis (e.g., x-axis) and a second axis
(e.g., y-axis), that allows the device to specify positions in a
plane.
[0066] According to one embodiment, the techniques herein are
performed by the computer system 200 in response to the processor
204 executing one or more sequences of one or more instructions
contained in the main memory 206. Such instructions are read into
the main memory 206 from another storage medium, such as the
storage device 210. Execution of the sequences of instructions
contained in the main memory 206 causes the processor 204 to
perform the process steps described herein. In alternative
embodiments, hard-wired circuitry is used in place of or in
combination with software instructions.
[0067] The term "storage media" as used herein refers to any
non-transitory media that store data and/or instructions that cause
a machine to operate in a specific fashion. Such storage media
includes non-volatile media and/or volatile media. Non-volatile
media includes, for example, optical disks, magnetic disks,
solid-state drives, or three-dimensional cross point memory, such
as the storage device 210. Volatile media includes dynamic memory,
such as the main memory 206. Common forms of storage media include,
for example, a floppy disk, a flexible disk, hard disk, solid-state
drive, magnetic tape, or any other magnetic data storage medium, a
CD-ROM, any other optical data storage medium, any physical medium
with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM,
NV-RAM, or any other memory chip or cartridge.
[0068] Storage media is distinct from but may be used in
conjunction with transmission media. Transmission media
participates in transferring information between storage media. For
example, transmission media includes coaxial cables, copper wire
and fiber optics, including the wires that comprise the bus 202.
Transmission media can also take the form of acoustic or light
waves, such as those generated during radio-wave and infrared data
communications.
[0069] In an embodiment, various forms of media are involved in
carrying one or more sequences of one or more instructions to the
processor 204 for execution. For example, the instructions are
initially carried on a magnetic disk or solid-state drive of a
remote computer. The remote computer loads the instructions into
its dynamic memory and send the instructions over a telephone line
using a modem. A modem local to the computer system 200 receives
the data on the telephone line and use an infrared transmitter to
convert the data to an infrared signal. An infrared detector
receives the data carried in the infrared signal and appropriate
circuitry places the data on the bus 202. The bus 202 carries the
data to the main memory 206, from which processor 204 retrieves and
executes the instructions. The instructions received by the main
memory 206 can optionally be stored on the storage device 210
either before or after execution by processor 204.
[0070] The computer system 200 also includes a communication
interface 218 coupled to the bus 202. The communication interface
218 provides a two-way data communication coupling to a network
link 220 that is connected to a local network 222. For example, the
communication interface 218 is an integrated service digital
network (ISDN) card, cable modem, satellite modem, or a modem to
provide a data communication connection to a corresponding type of
telephone line. As another example, the communication interface 218
is a local area network (LAN) card to provide a data communication
connection to a compatible LAN. In some implementations, wireless
links are also implemented. In any such implementation, the
communication interface 218 sends and receives electrical,
electromagnetic, or optical signals that carry digital data streams
representing various types of information.
[0071] The network link 220 typically provides data communication
through one or more networks to other data devices. For example,
the network link 220 provides a connection through the local
network 222 to a host computer 224 or to a cloud data center or
equipment operated by an Internet Service Provider (ISP) 226. The
ISP 226 in turn provides data communication services through the
world-wide packet data communication network now commonly referred
to as the "Internet" 228. The local network 222 and Internet 228
both use electrical, electromagnetic or optical signals that carry
digital data streams. The signals through the various networks and
the signals on the network link 220 and through the communication
interface 218, which carry the digital data to and from the
computer system 200, are example forms of transmission media. In an
embodiment, the network 220 contains the cloud or a part of the
cloud.
[0072] The computer system 200 sends messages and receives data,
including program code, through the network(s), the network link
220, and the communication interface 218. In an embodiment, the
computer system 200 receives code for processing. The received code
is executed by the processor 204 as it is received, and/or stored
in storage device 210, or other non-volatile storage for later
execution.
Autonomous Vehicle Architecture
[0073] FIG. 3 shows an example architecture 300 for an AV (e.g.,
the AV 100 shown in FIG. 1). The architecture 300 includes a
perception system 302 (sometimes referred to as a perception
circuit), a planning system 304 (sometimes referred to as a
planning circuit), a control system 306 (sometimes referred to as a
control circuit), a localization system 308 (sometimes referred to
as a localization circuit), and a database system 310 (sometimes
referred to as a database circuit). Each system plays a role in the
operation of the AV 100. Together, the systems 302, 304, 306, 308,
and 310 can be part of the AV system 120 shown in FIG. 1. In some
embodiments, any of the systems 302, 304, 306, 308, and 310 is a
combination of computer software (e.g., executable code stored on a
computer-readable medium) and computer hardware (e.g., one or more
microprocessors, microcontrollers, application-specific integrated
circuits [ASICs]), hardware memory devices, other types of
integrated circuits, other types of computer hardware, or a
combination of any or all of these things). Each of the systems
302, 304, 306, 308, and 310 is sometimes referred to as a
processing circuit (e.g., computer hardware, computer software, or
a combination of the two). A combination of any or all of the
systems 302, 304, 306, 308, and 310 is also an example of a
processing circuit.
[0074] In use, the planning system 304 receives data representing a
destination 312 and determines data representing a trajectory 314
(sometimes referred to as a route) that can be traveled by the AV
100 to reach (e.g., arrive at) the destination 312. In order for
the planning system 304 to determine the data representing the
trajectory 314, the planning system 304 receives data from the
perception system 302, the localization system 308, and the
database system 310.
[0075] The perception system 302 identifies nearby physical objects
using one or more sensors 121, e.g., as also shown in FIG. 1. The
objects are classified (e.g., grouped into types such as
pedestrian, bicycle, automobile, traffic sign, etc.) and a scene
description including the classified objects 316 is provided to the
planning system 304.
[0076] The planning system 304 also receives data representing the
AV position 318 from the localization system 308. The localization
system 308 determines the AV position by using data from the
sensors 121 and data from the database system 310 (e.g., a
geographic data) to calculate a position. For example, the
localization system 308 uses data from a GNSS (Global Navigation
Satellite System) sensor and geographic data to calculate a
longitude and latitude of the AV. In an embodiment, data used by
the localization system 308 includes high-precision maps of the
roadway geometric properties, maps describing road network
connectivity properties, maps describing roadway physical
properties (such as traffic speed, traffic volume, the number of
vehicular and cyclist traffic lanes, lane width, lane traffic
directions, or lane marker types and locations, or combinations of
them), and maps describing the spatial locations of road features
such as crosswalks, traffic signs or other travel signals of
various types. In an embodiment, the high-precision maps are
constructed by adding data through automatic or manual annotation
to low-precision maps.
[0077] The control system 306 receives the data representing the
trajectory 314 and the data representing the AV position 318 and
operates the control functions 320a-c (e.g., steering, throttling,
braking, and ignition) of the AV in a manner that will cause the AV
100 to travel the trajectory 314 to the destination 312. For
example, if the trajectory 314 includes a left turn, the control
system 306 will operate the control functions 320a-c in a manner
such that the steering angle of the steering function will cause
the AV 100 to turn left and the throttling and braking will cause
the AV 100 to pause and wait for passing pedestrians or vehicles
before the turn is made.
AV Inputs
[0078] FIG. 4 shows an example of inputs 402a-d (e.g., sensors 121
shown in FIG. 1) and outputs 404a-d (e.g., sensor data) that is
used by the perception system 302 (FIG. 3). One input 402a is a
LiDAR system (e.g., LiDAR 123 shown in FIG. 1). LiDAR is a
technology that uses light (e.g., bursts of light such as infrared
light) to obtain data about physical objects in its line of sight.
A LiDAR system produces LiDAR data as output 404a. For example,
LiDAR data is collections of 3D or 2D points (also known as a point
clouds) that are used to construct a representation of the
environment 190.
[0079] Another input 402b is a RADAR system. RADAR is a technology
that uses radio waves to obtain data about nearby physical objects.
RADARs can obtain data about objects not within the line of sight
of a LiDAR system. A RADAR system produces RADAR data as output
404b. For example, RADAR data are one or more radio frequency
electromagnetic signals that are used to construct a representation
of the environment 190.
[0080] Another input 402c is a camera system. A camera system uses
one or more cameras (e.g., digital cameras using a light sensor
such as a charge-coupled device [CCD]) to obtain information about
nearby physical objects. A camera system produces camera data as
output 404c. Camera data often takes the form of image data (e.g.,
data in an image data format such as RAW, JPEG, PNG, etc.). In some
examples, the camera system has multiple independent cameras, e.g.,
for the purpose of stereopsis (stereo vision), which enables the
camera system to perceive depth. Although the objects perceived by
the camera system are described here as "nearby," this is relative
to the AV. In some embodiments, the camera system is configured to
"see" objects far, e.g., up to a kilometer or more ahead of the AV.
Accordingly, in some embodiments, the camera system has features
such as sensors and lenses that are optimized for perceiving
objects that are far away.
[0081] Another input 402d is a traffic light detection (TLD)
system. A TLD system uses one or more cameras to obtain information
about traffic lights, street signs, and other physical objects that
provide visual navigation information. A TLD system produces TLD
data as output 404d. TLD data often takes the form of image data
(e.g., data in an image data format such as RAW, JPEG, PNG, etc.).
A TLD system differs from a system incorporating a camera in that a
TLD system uses a camera with a wide field of view (e.g., using a
wide-angle lens or a fish-eye lens) in order to obtain information
about as many physical objects providing visual navigation
information as possible, so that the AV 100 has access to all
relevant navigation information provided by these objects. For
example, the viewing angle of the TLD system is about 120 degrees
or more.
[0082] In some embodiments, outputs 404a-d are combined using a
sensor fusion technique. Thus, either the individual outputs 404a-d
are provided to other systems of the AV 100 (e.g., provided to a
planning system 304 as shown in FIG. 3), or the combined output can
be provided to the other systems, either in the form of a single
combined output or multiple combined outputs of the same type
(e.g., using the same combination technique or combining the same
outputs or both) or different types type (e.g., using different
respective combination techniques or combining different respective
outputs or both). In some embodiments, an early fusion technique is
used. An early fusion technique is characterized by combining
outputs before one or more data processing steps are applied to the
combined output. In some embodiments, a late fusion technique is
used. A late fusion technique is characterized by combining outputs
after one or more data processing steps are applied to the
individual outputs.
[0083] FIG. 5 shows an example of a LiDAR system 502 (e.g., the
input 402a shown in FIG. 4). The LiDAR system 502 emits light
504a-c from a light emitter 506 (e.g., a laser transmitter). Light
emitted by a LiDAR system is typically not in the visible spectrum;
for example, infrared light is often used. Some of the light 504b
emitted encounters a physical object 508 (e.g., a vehicle) and
reflects back to the LiDAR system 502. (Light emitted from a LiDAR
system typically does not penetrate physical objects, e.g.,
physical objects in solid form.) The LiDAR system 502 also has one
or more light detectors 510, which detect the reflected light. In
an embodiment, one or more data processing systems associated with
the LiDAR system generates an image 512 representing the field of
view 514 of the LiDAR system. The image 512 includes information
that represents the boundaries 516 of a physical object 508. In
this way, the image 512 is used to determine the boundaries 516 of
one or more physical objects near an AV.
[0084] FIG. 6 shows the LiDAR system 502 in operation. In the
scenario shown in this figure, the AV 100 receives both camera
system output 404c in the form of an image 602 and LiDAR system
output 404a in the form of LiDAR data points 604. In use, the data
processing systems of the AV 100 compares the image 602 to the data
points 604. In particular, a physical object 606 identified in the
image 602 is also identified among the data points 604. In this
way, the AV 100 perceives the boundaries of the physical object
based on the contour and density of the data points 604.
[0085] FIG. 7 shows the operation of the LiDAR system 502 in
additional detail. As described above, the AV 100 detects the
boundary of a physical object based on characteristics of the data
points detected by the LiDAR system 502. As shown in FIG. 7, a flat
object, such as the ground 702, will reflect light 704a-d emitted
from a LiDAR system 502 in a consistent manner. Put another way,
because the LiDAR system 502 emits light using consistent spacing,
the ground 702 will reflect light back to the LiDAR system 502 with
the same consistent spacing. As the AV 100 travels over the ground
702, the LiDAR system 502 will continue to detect light reflected
by the next valid ground point 706 if nothing is obstructing the
road. However, if an object 708 obstructs the road, light 704e-f
emitted by the LiDAR system 502 will be reflected from points
710a-b in a manner inconsistent with the expected consistent
manner. From this information, the AV 100 can determine that the
object 708 is present.
Ghost Point Occurrence
[0086] As previously noted, the term "ghost point" refers to a
point in a LiDAR point cloud based on erroneous return points in
the point cloud that is captured during a LiDAR scan. More
specifically, ghost points are the result of horizontal diffusion
of the LiDAR beam during the transmit and return path (e.g., the
path from the LiDAR emitter to an object, and then the reflection
from the object to the LiDAR receiver) leading to multipath
effects. The presence of ghost points will introduce inaccuracies
into the point cloud, which may negatively impact actions of an AV
that relies upon the point cloud data for functions such as
autonomous navigation.
[0087] FIG. 8 shows an example comparison between a real-world
image and a point cloud which may result from a ToF system.
Specifically, FIG. 8 depicts an example of a real-world image 805
of a scenario and the resultant model 800 of the scenario. In the
real-world image 805, an individual 810 is adjacent to, but
separate from, a sign 815. The point clouds 820, 825 of the
individual 810 and the sign 815, respectively, are depicted in the
model at 800.
[0088] It will be understood that in some embodiments, the point
clouds 820 and 825 may be considered as different regions or zones
of a single point cloud, or a merged point cloud. However, for the
sake of ease of description here, elements 820 and 825 will be
referred to as separate point clouds. In one embodiment, the point
clouds 820 and 825 may be point clouds produced by a single LiDAR
system, while in other embodiments the point clouds 820 and 825 may
be point clouds produced by a separate LiDAR systems.
[0089] As may be seen in FIG. 8, a number of points 830 are
positioned between the point clouds 820, 825 of the individual and
the sign, respectively. Additionally, points 835 are present. By
comparison of the model 800 with the real-world image 805, it may
be seen that there is no object between the individual 810 and the
sign 815. Rather, the points that make up point cloud 830 are ghost
points that are the result of horizontal diffusion of reflections
generated as a result of operation of the LiDAR system that
generated the point clouds 820 and 825. Specifically, the point
cloud 830 may be a result of diffusion based on one or both of a
retroreflector such as the sign 815 or some other retroreflector.
As used herein, a retroreflector is a device or surface that
reflects radiation (usually light) back to its source with minimum
scattering and minimum attenuation.
[0090] Additionally, it may be seen that there is no object to the
right of the sign 815 in the real-world image 805. Rather, the
points that make up point cloud 835 are ghost points that are the
result of horizontal diffusion of reflections generated as a result
of operation of the LiDAR system. Similarly to point cloud 830, the
point cloud 835 may be a result of diffusion based on the sign
815.
[0091] Similarly to elements 820 and 825, in another embodiment the
point clouds 830 and 835 may be referred to as a region or zone of
a single larger point cloud. However, for the sake of description
herein, element 830 will be referred to as an individual point
cloud.
[0092] In one scenario, the point cloud 830 would be interpreted by
an object detector (e.g., a deep neural network) of a perception
system (e.g., perception system 302 of AV 100) as a solid object,
and so the AV (e.g., the planning system 304) would attempt to
navigate around the object. Additionally, because the point cloud
830 is based on diffusion reflections (or, alternatively,
"multipath returns") generated as a result of operation of the
LiDAR system, as the LiDAR system moves with the AV and resamples
the environment, the point cloud 830 may change or disappear, which
could further impact the accuracy of the object detector and thus
the operation of the AV.
[0093] FIGS. 9a and 9b (collectively referred to as "FIG. 9")
depict an example transmission structure and receive structure of a
ToF system such as a LiDAR system. Specifically, FIG. 9a depicts a
simplified example of a transmission structure (e.g., a
"transmitter") of a LiDAR system, and FIG. 9b depicts a simplified
example of a receive structure (e.g., a "receiver") of the LiDAR
system.
[0094] The example transmitter includes an emitter 905. In the
embodiment of a LiDAR system, the emitter 905 is an optical emitter
(e.g., a laser diode) that is configured to emit a light beam 910
in the visible spectrum, the infrared spectrum, the ultraviolet
spectrum, or some other spectrum. In another embodiment, the
emitter 905 is a different type of emitter which emits a signal in
the radio spectrum or some other type of electromagnetic
signal.
[0095] In the example transmitter shown, cavity 915 includes
reflective surfaces that reflect the light beam 910 as it
propagates the length of the cavity 915 until it reaches a lens
920. The lens 920 is configured to focus the light beam 910 along a
pre-defined field of view 935. The transmitter also includes an
optical dome 925 which, in an embodiment, is a transparent material
that is arranged to generally cover the lens 920 to protect the
lens from damage such as a scratch. The optical dome 925 is, in an
embodiment, curved and as may be seen in FIG. 9a, causes a
reflection of the light beam 910, which is then reflected by the
lens 920 to exit the optical dome 925 as reflected light beams 923.
One or more of the reflected light beams 923 impinge upon a
reflective surface such as a retroreflector 930, which reflects the
light beams back to receiver
[0096] Referring to FIG. 9b, the receiver receives reflected light
beams 945 and 950 which are refracted and focused by a lens 920
into cavity 915, where they propagate through cavity 915 until they
arrive at a photodiode 940. The photodiode 940 is configured to
register the arrival of an optical signal and convert the optical
signal into an electrical signal which is then supplied to another
element of the LiDAR system (or an electrical device
communicatively coupled with the LiDAR system). In an embodiment,
photodiode 940 is an avalanche photodiode (APD).
[0097] In the embodiment depicted in FIG. 9, the transmission field
of view 935 and the optical reception path 945 are considered to be
the nominal directions. That is, a processor coupled with the APD
940 expects that an optical signal received by the APD 940 will
arrive in the direction indicated by optical path 945. However,
because of the retroreflectors 930, reflected light beams 950 that
are different from the nominal direction indicated by the optical
path 945 will be received. These reflected light beams 950 will be
highly attenuated, and therefore have a low optical intensity. In
this situation, the reflected light beams received along optical
paths 950 will register as low-intensity points received from the
nominal direction (e.g., along the optical path 945), resulting in
"ghost points" as described above.
[0098] Often, ghost points will appear on either side of a
retroreflector such as a street sign or some other object, or as an
aura around the object. The ghost points will have a similar shape
and physical structure as the retroreflector, and appear at the
same range from the LiDAR system as the retroreflector itself.
Ghost Point Filter Techniques
[0099] A ghost point is based on a reflection from a retroreflector
(e.g., a street sign, a vehicle, etc. as described above) that is
angled with respect to the nominal direction of the LiDAR system.
In an embodiment, the angle at which the presence of a
retroreflector will cause a ghost point is a fixed angle for a
given system which, for a LiDAR system, is related to the optical
design. For example, as a LiDAR system (e.g., a LiDAR system that
may be the same as or similar to LiDAR system 502, above)
approaches a retroreflector (e.g., attached to the roof of an AV
100), the observed ghost point may move closer to the
retroreflector or, in certain cases, disappear from the LiDAR scan
depending on how the material and shape of the retroreflector which
determines how the light will reflect.
[0100] Based on this fixed angle, a "ghost zone" is broadly defined
as a region at a fixed angle from the LiDAR system that leads to
the presence of a ghost point, as described fully in reference to
FIGS. 10 and 11.
[0101] FIG. 10 depicts an example of ghost point regions.
Specifically, FIG. 10 depicts an example of ghost point regions
1005 and 1010 on either side of a center point 1000 (defined below)
based on a calibration test conducted to identify the location of
said ghost point regions 1005/1010.
[0102] To conduct the calibration test, in this embodiment, a LiDAR
system is fixated and the location of the nominal direction is
marked with the center point 1000. In other words, the location of
the center point 1000 corresponds to the nominal optical path 945.
The center point 1000 is a point on a test panel that includes a
plurality of LEDs. Respective LEDs of the test panel are then
activated. LEDs in the ghost point regions 1005/1010 will generate
light beams that are analogous to the reflected light beams 950
received by the photodiode described above in reference to FIG. 9.
That is, LEDs in the ghost point regions 1005/1010 will cause the
receiver of the actual LiDAR system to register a low-intensity
optical signal although they are outside of an intended nominal
transmission and reception path of the LiDAR system.
[0103] Through this calibration test procedure, ghost point regions
1005/1010 are identified for a particular center point 1000. In one
embodiment, a single ghost point region or pair of ghost point
regions (e.g., ghost point regions 1005/1010) are used for a LiDAR
system. In another embodiment, different ghost point regions are
defined for different locations of the center point 1000. That is,
a different calibrated ghost point region is used for different
elevation angles (e.g., scan lines) of the LiDAR scan or different
azimuth (e.g., lateral) locations.
[0104] To apply a ghost point removal filter based on the
identified ghost point regions, a range image is identified and
stored as a tensor. Specifically, the LiDAR system generates a
range image that includes a plurality of LiDAR return points. FIG.
11 depicts an example of a range image for processing by a ghost
point removal filter.
[0105] As shown at 1100, the range image includes a number of data
points 1110. Each of the data points 1110 has data related to an
elevation angle (which, in one embodiment, 0.60 degrees, 0.26
degrees, and -0.08 degrees) as measured from a direction in which
the LiDAR system is pointing when the range image is generated.
Additionally, each of the data points 1110 has an azimuth angle
(which, in one embodiment, is -0.4 degrees, 0 degrees, 0.4 degrees,
and 0.8 degrees) as measured from a direction in which the LiDAR
system is pointing when the data points are collected. It will be
understood that these example angles are intended only as examples
related to one configuration of a tensor, and another embodiment
may include different values for the elevation or azimuth angles,
or different separations between each of the data points 1110.
[0106] The data points 1110 are then organized as a tensor 1105,
which includes a number of rows, columns, and channels.
Specifically, the rows and columns correspond to the elevation
angles and azimuth angles, respectively, as shown in 1100. The
tensor 1105 then includes a number of channels 1115, 1120, and
1125, which correspond to different values related to each of the
data points 1110. For example, one channel 1125 corresponds to
range information related to each of the data points 1110 (e.g.,
the measured distance of the object that reflected the optical
signal and generated the data point 1110). Another channel 1120
corresponds to the intensity of the data point 1110, that is, the
received energy of the optical signal. Another channel 1115
corresponds to the point index of the data point 1110, that is, an
identifier of the data point within the range image. These channels
are, however, an example in accordance with one embodiment, and
another embodiment may have more or fewer channels, or channels
arranged in a different configuration within the tensor 1105.
[0107] After the range image has been converted to a tensor such as
tensor 1105, and the ghost point regions have been identified, a
kernel is constructed to filter low-intensity ghost points from the
range image. Generally, a kernel is a matrix that is applied to an
image (or, more specifically, the tensor 1105) to identify and
alter or extract certain features (e.g., ghost points) from the
image. In one embodiment, the kernel is a convolutional kernel, a
convolutional neural network (CNN) kernel, a machine learning
(ML)-based kernel, etc. In embodiments where the kernel is a CNN
kernel or a ML-based kernel, the kernel may be based on or modified
by, for example, the specific ghost point zones that are identified
or used, the number of ghost points that are identified or filtered
based on use of the kernel, the size or shape of the
retroreflector, etc.
[0108] FIG. 12 depicts a high-level example technique of removal of
a ghost point from a range image based on a kernel. In an
embodiment, the technique is performed by, in whole or in part, a
perception system such as perception system 302 of FIG. 3, a
processor such as processor 204 of FIG. 2, or some other element of
a LiDAR system or vehicle as described herein.
[0109] Initially, the technique includes finding all low-intensity
points in the range image at 1205. Specifically, the technique
includes finding low-intensity points in the tensor 1105.
Identifying the low-intensity points, in one embodiment, includes
identifying data points in the tensor with an intensity value (e.g.
a value of the intensity channel 1120 of a data point 1110) at or
below a threshold. In one embodiment, the threshold is
pre-identified based on, for example, the type of LiDAR system used
or some other characteristic of the LiDAR system. As one example,
the threshold is pre-identified during the calibration process of
the LiDAR system as described with respect to FIG. 10. In another
embodiment, the threshold is identified based on intensity values
of the range image. That is, the threshold is identified through
analysis of the intensity values of various data points in the
range image. As an example of this embodiment, the threshold is
dynamic (e.g., the threshold changes over time based on the range
image or some other factor). In another embodiment, the threshold
is identified based on one or more additional or alternative
factors.
[0110] The technique further includes selection of an identified
low-intensity point to test at 1210. Specifically, one of the
low-intensity points identified at 1205 is then selected for
review.
[0111] The technique further includes identifying, at 1215, a
filtering kernel based on the selected low-intensity point. As
previously noted, in one embodiment a single filtering kernel is
used for the entire tensor 1105. In another embodiment, different
filtering kernels are used based on the selected low-intensity
point. For example, different filtering kernels are used based on
different elevation angles, azimuth angles, etc.
[0112] The technique further includes finding, at 1220, all
high-intensity points that fall within the kernel-based ghost
zone(s). Specifically, and as will be explained in greater detail
with respect to FIG. 13, application of the filtering kernel
includes overlaying the filtering kernel on the tensor 1105. The
filtering kernel is based on the ghost point regions 1005/1010 and
the center point 1000 described with respect to FIG. 10.
Conceptually, and at a high-level for the sake of description, the
center point 1000 is aligned with the low-intensity point selected
at 1210. A number of high-intensity points (e.g., data points 1110
with a value of the intensity channel 1120 at or above a threshold)
is then identified at 1220. Similarly to the low-intensity points
described above, in one embodiment the high-intensity points are
identified based on a pre-identified threshold. In another
embodiment, the threshold is identified based on analysis of the
intensity values of various data points in the range image and, as
a particular example, is dynamic as described above. In yet another
embodiment, the threshold is identified based on one or more
additional or alternative factors.
[0113] In one embodiment, the low-intensity threshold and the
high-intensity threshold are separated from one another such that
the low-intensity threshold has optical intensity values between 0
and x, and the high-intensity threshold has optical intensity
values between y and 255. Values between x and y may be considered
to be neither low-intensity values nor high-intensity values and
therefore are not identified at either 1205 or 1220. In another
embodiment, the low-intensity values are between 0 and x, and the
high-intensity values are between x and 255.
[0114] The number of high-intensity points in the ghost zone(s) are
then counted at 1225. In one embodiment, only high-intensity points
in a single ghost zone (e.g., only one of ghost zones 1005 or 1010)
are counted. In another embodiment, high-intensity points in two or
more ghost zones (e.g., ghost zones 1005 and 1010) are counted.
[0115] In one embodiment, a distance to the low-intensity point
identified at 1205 is determined. Specifically, the distance to the
low-intensity point from the vehicle or LiDAR system is determined.
In this embodiment, only high-intensity points that have a distance
within a range of the distance to the low-intensity point may be
identified at 1220 or counted at 1225. In this manner, the system
is able to verify that the low-intensity points and the
high-intensity points are related to one another such that the
low-intensity point may be a ghost point that is caused by the
high-intensity point rather than data points at two different
distances, which would mean that they are likely unrelated to one
another. In one embodiment, the range of the distance may be on the
order of +/-1 meter (m), while in other embodiments the range of
the distance may be on the order of +/-0.5 m, +/-0.25 m, etc. The
specific range of the distance may be different based on, for
example, the type of LiDAR system used, the initial distance to the
low-intensity point identified at 1205, etc.
[0116] The number of high-intensity points is then compared against
a region number threshold at 1230. Specifically, the number of
high-intensity points identified within the ghost zone(s) at 1225
are compared against a value associated with the region number
threshold. If the number of high-intensity points is above (or,
optionally, at or above) the value associated with the region
number threshold, then the low-intensity point selected at 1210 is
identified as a ghost point and removed from the range image on
which the tensor 1105 is based, or more specifically, the point
cloud of the range image. If the number of high-intensity points is
below (or, optionally, at or below) the value associated with the
region number threshold, then the low-intensity point is identified
to not be a ghost point and is not removed from the range image.
Similarly to the other thresholds described above, in one
embodiment the region number threshold is based on a pre-identified
value related to a factor such as the type of LiDAR system (e.g.,
one or more components of the LiDAR system), the environment in
which the LiDAR system is located, etc. In another embodiment, the
region number threshold is identified based on analysis of the
range image or some other factor. The technique may then return to
1210 to select another low-intensity point of the low-intensity
points identified at 1205.
[0117] It will be understood that this technique is intended as an
example technique, and another embodiment will include a technique
with one or more variations from that depicted in FIG. 12. For
example, another embodiment may have more or fewer elements than
depicted, elements in a different order than depicted, etc.
[0118] FIG. 13 depicts a graphical example of the identification
and removal of a ghost point based on a ghost point region.
Specifically, FIG. 13 depicts a range image 1325 that includes
elements that are similar to those depicted in model 800 of FIG. 8.
The range image 1325 depicts a point cloud 1305 related to an
individual, a point cloud 1307 related to a sign, and a point cloud
1300 of low-intensity points which are respectively similar to the
point cloud 820 of the individual 810, point cloud 825 of the sign
815, and the point cloud 830.
[0119] FIG. 13 further depicts the overlay of a filtering kernel
onto the range image 1325. Specifically, FIG. 13 depicts a
low-intensity point 1315 that is selected for analysis, for example
as described above with respect to element 1210. Based on that
low-intensity point, a filtering kernel is identified as described
with respect to element 1215. The filtering kernel includes ghost
zones 1310a and 1310b, which are respectively similar to ghost
zones 1005 and 1010 of FIG. 10.
[0120] The point clouds 1305 and 1307 both include high-intensity
points. As described with respect to elements 1220 and 1225, the
high-intensity points in the ghost zones 1310a and 1310b are
identified and counted. For example, the high-intensity points 1320
corresponding to the point cloud 1307 related to the sign within
the ghost zone 1310b are identified and counted. Additionally or
alternatively, the high-intensity points 1330 (if any) of the point
cloud 1305 related to the individual within the ghost zone 1310a
are identified and counted. The number of high-intensity points are
then compared to a value associated with a region number threshold
as described with respect to element 1230. If the number of
high-intensity points are above (or, alternatively, at or above)
the value associated with the region number threshold, then the
low-intensity point 1315 is identified as a ghost point and removed
from the point cloud 1300 and the range image 1325.
[0121] FIG. 14 depicts an example of a LiDAR pipeline 1400.
Typically, the LiDAR pipeline 1400 is enacted by, or an element of,
a system such as the perception system 302 of FIG. 3. More
generally, the LiDAR pipeline 1400 is enacted by, or an element of,
a processor such as processor 204 of FIG. 2.
[0122] Initially, a number of LiDAR range images 1405a and 1405b
are input to the pipeline 1400. It will be understood that although
only two LiDAR range images are depicted in FIG. 14, another
embodiment includes more or fewer range images 1405a/1405b.
Respective ones of the LiDAR range images 1405a/1405b are similar
to the range image 1325 described above and include various point
clouds such as point clouds 1300, 1305, and 1307.
[0123] The range images 1405a/1405b are input to ghost point
filters 1410a/1410b. Respective ones of the ghost point filters
1410a/1410b apply a filtering kernel to identify and remove ghost
points, as described with respect to FIG. 12. As shown in FIG. 14,
the ghost point filters 1410a and 1410b are independent from one
another. For example, the ghost point filters 1410a and 1410b may
be enacted by different hardware, software, firmware, etc. In
another embodiment, the ghost point filters 1410a and 1410b may be
the same filter, or subelements of the same filter.
[0124] The range images, after ghost point filtering, are then
provided to other elements of the LiDAR pipeline. For example, in
one embodiment they are output and provided to a foreground
extraction system 1410 where the images are merged and processed.
The foreground extraction system 1415, in one embodiment, is
configured to extract foreground objects in the range image. The
output of the foreground extraction system 1415 includes the
extracted foreground objects, and is provided to a segmentation
system which is configured to classify point clouds into homogenous
regions that have similar properties to one another. The output of
the segmentation system 1420 includes an indication of the
classified point clouds, and, more particularly, individually
classified point clouds. The classified point clouds are provided
to a noise filter 1425 which is configured to remove noise from the
point clouds (e.g., artifacts from errors or other data points that
do not correspond to one or more of the homogenous regions from the
segmentation system).
[0125] It will be understood that this pipeline is intended as an
example pipeline in accordance with one embodiment, and other
embodiments will include more or fewer elements, or elements
arranged in a different order. However, it will be further
understood that the presence of the ghost point filters 1410a/1410b
at an initial stage of the LiDAR pipeline will provide significant
benefits in that the filters will operate on respective LiDAR point
clouds independently, and, as a result, be computationally
efficient.
[0126] FIG. 15 depicts an example of a technique related to
identifying and removing a ghost point from a point cloud.
Generally, the technique of FIG. 15 may be considered to be similar
to the technique of FIG. 12. Similarly to the technique of FIG. 12
the technique is performed by, in whole or in part, a perception
system such as perception system 302 of FIG. 3, a processor such as
processor 204 of FIG. 2, or some other element of a LiDAR system or
vehicle as described herein.
[0127] The technique includes obtaining, at 1505, a range image
from a depth sensor of a vehicle operating in an environment. The
range image is similar to, for example, range image 1325 or some
other range image described herein.
[0128] The technique further includes identifying, at 1510, a first
data point in the range image with an intensity at or below a first
intensity threshold. This data point is, for example, the
low-intensity data point described with respect to element 1210 of
FIG. 12. The first intensity threshold is the intensity threshold
described with respect to, for example, element 1205 of FIG.
12.
[0129] The technique further includes determining, at 1515, a
number of data points in the range image that have an intensity at
or above a second intensity threshold in a first region of the
range image. Specifically, the technique includes determining the
number of high-intensity data points in a region of the range image
such as the ghost zones described with respect to, for example,
elements 1005/1010 or 1310a/1310b. Determination of the number of
high-intensity data points in these regions is as described with
respect to, for example, elements 1220 and 1225.
[0130] The technique further includes determining, at 1520, if the
number of data points identified at 1515 is at or above a region
number threshold. This determination and the region number
threshold are similar to, for example, the determination and
threshold value described with respect to element 1230.
[0131] The technique further includes removing, at 1525, the first
data point from the range image if the number of data points is at
or above the region number threshold. For example, as described
with respect to element 1230, the first data point (e.g., the
low-intensity data point) is removed from the range image if the
number of identified high-intensity data points within the ghost
zone(s) is at or above a region number threshold. This is because
the low-intensity data point is identified as being a ghost point
based on the presence of the high-intensity data points within the
ghost zones.
[0132] The technique further includes operating (or facilitating
operation of), at 1530, of the vehicle in the environment based at
least in part on the range image. For example, in an embodiment the
processor then uses the output range image as the basis for
functions such as path planning of the vehicle. In another
embodiment, the processor outputs the range image to another
processor or system of the vehicle (e.g., control system 306 of
FIG. 3) which then uses the range image to control one or more
functions of the vehicle.
[0133] Similarly to FIG. 12 described above, it will be recognized
that the technique of FIG. 15 is intended as an example technique
in accordance with one embodiment, and another embodiment may vary.
For example, another embodiment will include more or fewer elements
than those depicted in FIG. 15, elements arranged in a different
order than depicted, etc.
[0134] Other variations may be present in yet further embodiments.
For example, one embodiment includes an additional element related
to distinguishing different point clouds that are produced by
different LiDAR systems to identify which point cloud is the cause
of which ghost points. In another embodiment, the first data point
(e.g., the ghost point) is not removed from the image, but rather
information related to the first data point is altered. For
example, a flag value is added or altered to indicate that the
first data point is identified as a "ghost point." In this
embodiment, the first data point remains in the image, but is
processed differently by a downstream component of a LiDAR pipeline
such as pipeline 1400.
[0135] In the foregoing description, embodiments of the invention
have been described with reference to numerous specific details
that may vary from implementation to implementation. The
description and drawings are, accordingly, to be regarded in an
illustrative rather than a restrictive sense. The sole and
exclusive indicator of the scope of the invention, and what is
intended by the applicants to be the scope of the invention, is the
literal and equivalent scope of the set of claims that issue from
this application, in the specific form in which such claims issue,
including any subsequent correction. Any definitions expressly set
forth herein for terms contained in such claims shall govern the
meaning of such terms as used in the claims. In addition, when we
use the term "further comprising," in the foregoing description or
following claims, what follows this phrase can be an additional
step or entity, or a sub-step/sub-entity of a previously-recited
step or entity.
* * * * *