Obstacle filtering system based on point cloud features

Abstract

Methods, apparatuses, and systems for filtering candidate obstacles determined based on output of a LIDAR device in an autonomous vehicle are disclosed. A point cloud comprising a plurality of points is generated based on output of the LIDAR device (610). One or more candidate obstacles are determined based on the point cloud (620). The one or more candidate obstacles are filtered to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise based at least in part on a characteristic associated with points corresponding to each of the one or more candidate obstacles (630). One or more identified obstacles comprising candidate obstacles that have not been removed are determined (640). Operation of the autonomous vehicle is controlled based on the identified obstacles (650).

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure generally relate to operating an autonomous vehicle. More specifically, embodiments of the present disclosure relate to processing candidate obstacles determined based on outputs of LIDAR devices in an autonomous vehicle. BACKGROUND

[0002] A vehicle operating in an autonomous mode (e.g., driverless) can free up passengers, especially drivers, from some driving-related duties. When operating in an autonomous mode, a vehicle can use on-board sensors to navigate to various locations, allowing the vehicle to travel with minimal human interaction or in some cases, without any passengers.

[0003] LIDAR (light detection and ranging, or a hybrid of light and radar) technology has been widely used in military, geography, oceanography, and in autonomous vehicles in the last decade. LIDAR devices can estimate distances to objects when scanning through a scene to assemble a point cloud representing the surfaces of objects that reflect light. Each point in the point cloud can be determined by emitting a pulse of laser light and detecting a return pulse reflected from an object, if any, and determining the distance to the object based on the time delay between the emitted pulse and the received reflection. A beam or beams of laser light can be rapidly repeated across a scene to provide continuous real-time information of distances to reflecting objects in the scene. One revolution of a beam of laser light produces one revolution of points.

[0004] LIDAR technology is susceptible to noise, such as caused by dust or other particles. Noisy point clouds can cause perception modules to generate false positives of obstacles. SUMMARY

[0005] Embodiments of the present disclosure provide computer-implemented methods, non-transitory machine-readable media, and data processing systems.

[0006] Some embodiments of the present disclosure provide a computer-implemented method, comprising: generating, based on outputs of light detection and ranging (LIDAR) devices, a point cloud comprising a plurality of points; determining one or more candidate obstacles based on the point cloud; filtering the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise based at least in part on features associated with points corresponding to each of the one or more candidate obstacles; determining one or more identified obstacles comprising candidate obstacles that have not been removed; and controlling operation of an autonomous vehicle based on the identified obstacles.

[0007] Some implementations of the present disclosure provide a non-transitory machine-readable medium having stored therein instructions that, when executed by a processor, cause the processor to perform operations comprising: generating, based on an output of a light detection and ranging (LIDAR) device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles that correspond to noise based at least in part on a feature associated with points corresponding to each of the one or more candidate obstacles; determining one or more identified obstacles comprising candidate obstacles that have not been removed; controlling operation of an autonomous vehicle based on the identified obstacles.

[0008] Some implementations of the present disclosure provide a data processing system comprising: a processor; and a memory coupled to the processor to store instructions that, when executed by the processor, cause the processor to perform operations comprising: generating, based on an output of a light detection and ranging (LIDAR) device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles that correspond to noise based at least in part on a feature associated with points corresponding to each of the one or more candidate obstacles; determining one or more identified obstacles comprising candidate obstacles that have not been removed; controlling operation of an autonomous vehicle based on the identified obstacles.

[0009] Some implementations of the present disclosure provide a computer program product comprising a computer program which, when executed by a processor, implements a method comprising: generating, based on an output of a light detection and ranging (LIDAR) device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles that correspond to noise based at least in part on a feature associated with points corresponding to each of the one or more candidate obstacles; determining one or more identified obstacles comprising candidate obstacles that have not been removed; controlling operation of an autonomous vehicle based on the identified obstacles. BRIEF DESCRIPTION OF DRAWINGS

[0010] Implementations of the present disclosure are illustrated by way of example, and not by way of limitation, in the accompanying drawings, in which like reference numerals indicate similar elements.

[0011] Figure 1 is a block diagram illustrating a networked system, in accordance with one implementation.

[0012] Figure 2 is a block diagram illustrating an example of an autonomous vehicle, in accordance with one implementation.

[0013] Figures 3A-3B is a block diagram illustrating an example of a perception and planning system for use with an autonomous vehicle, according to one embodiment.

[0014] Figure 4 is a block diagram illustrating various modules available, according to one embodiment.

[0015] Figure 5A and Figure 5B is a graph illustrating a spatial distribution of LIDAR points corresponding to an obstacle, according to one embodiment.

[0016] Figure 6 is a flowchart illustrating an example method for filtering candidate obstacles determined based on output of a LIDAR device in an autonomous vehicle, according to one embodiment.

[0017] Figure 7 is a flowchart illustrating an example method for filtering candidate obstacles determined based on output of a LIDAR device in an autonomous vehicle, according to one embodiment. DETAILED DESCRIPTION

[0018] Various embodiments and aspects of the disclosure will be described with reference to details discussed in the following description and illustrated in the accompanying drawings. The following description and drawings are illustrative of the disclosure and are not to be construed as limiting the disclosure. Numerous specific details are described to provide a thorough understanding of various embodiments of the present disclosure. However, in certain instances, well-known or conventional details are not described in order to provide a concise discussion of embodiments of the present disclosure.

[0019] Reference in the specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the disclosure. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

[0020] According to some embodiments, methods, apparatuses, and systems for filtering candidate obstacles determined based on output of a LIDAR device in an autonomous vehicle are disclosed. Based on the output of the LIDAR device, a point cloud comprising a plurality of points is generated. Based on the point cloud, one or more candidate obstacles are determined. The one or more candidate obstacles are filtered to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise based at least in part on a feature associated with points corresponding to each of the one or more candidate obstacles. One or more identified obstacles comprising candidate obstacles that have not been removed are determined. Based on the identified obstacles, operations of the autonomous vehicle are controlled.

[0021] In one embodiment, the first set of candidate obstacles corresponding to noise includes candidate obstacles corresponding to points corresponding to dust. In one embodiment, filtering the features associated with the points corresponding to each of the one or more candidate obstacles on which the one or more candidate obstacles are based includes one or more of a distribution of intensity measurements of the points corresponding to each of the one or more candidate obstacles, a spatial distribution of the points corresponding to each of the one or more candidate obstacles, or a combination thereof.

[0022] In one embodiment, filtering the one or more candidate obstacles includes removing each candidate obstacle corresponding to a point in which a ratio of a number of points associated with low intensity measurements to a total number of points corresponding to the candidate obstacle is higher than a first threshold. A point is associated with low intensity measurements when an intensity measurement of the point is lower than an intensity threshold.

[0023] In one embodiment, for each remaining candidate obstacle for which the identified obstacle type is a vehicle or for which the identified physical dimension is greater than a dimension threshold, projecting the points corresponding to the candidate obstacle to a first dimension and a second dimension of a horizontal plane. The first dimension and the second dimension are orthogonal to each other. Determining a first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension. Removing the candidate obstacle when the first standard deviation is higher than a first standard deviation threshold or when the second standard deviation is higher than a second standard deviation threshold. In one embodiment, the candidate obstacle is removed when the first standard deviation is higher than the first standard deviation threshold and the second standard deviation is higher than the second standard deviation threshold.

[0024] In one embodiment, the first standard deviation threshold is equal to the second standard deviation threshold. In another embodiment, the first standard deviation threshold is different from the second standard deviation threshold.

[0025] In one embodiment, for each remaining candidate obstacle for which the identified obstacle type is a vehicle or for which the identified physical dimension is greater than a dimension threshold, projecting the points corresponding to the candidate obstacle to an area of a horizontal plane. The area of the horizontal plane is associated with a first number of grids. Determining a second number of grids within the first number of grids, each second grid containing at least one projected point. Removing the candidate obstacle when a ratio of the second number of grids to the first number of grids is higher than a second threshold.

[0026] In one embodiment, filtering the one or more candidate obstacles further includes removing each candidate obstacle corresponding to a point in which a ratio of a number of points associated with low height to a total number of points corresponding to the candidate obstacle is lower than a third threshold. A point is associated with low height when a height of the point is lower than a height threshold.

[0027] In one embodiment, filtering the one or more candidate obstacles further includes removing each candidate obstacle that is not present in any perception result of a predetermined number of immediately preceding perception cycles.

[0028] Figure 1 is a block diagram illustrating an autonomous vehicle network configuration according to one embodiment of the present disclosure. Referring to Figure 1 , the network configuration 100 includes an autonomous vehicle 101 communicatively coupled to one or more servers 103-104 through a network 102. Although one autonomous vehicle is shown, multiple autonomous vehicles can be coupled to each other and / or to the servers 103-104 through the network 102. The network 102 can be any type of network, for example, a wired or wireless local area network (LAN), a wide area network (WAN) such as the Internet, a cellular network, a satellite network, or a combination thereof. The servers 103-104 can be any type of server or server cluster, such as a web or cloud server, an application server, a backend server, or a combination thereof. The servers 103-104 can be a data analytics server, a content server, a traffic information server, a map and point of interest (MPOI) server, or a location server, among others.

[0029] An autonomous vehicle refers to a vehicle that can be configured to be in an autonomous mode in which the vehicle navigates through an environment with little or no input from a driver. Such an autonomous vehicle can include a sensor system having one or more sensors configured to detect information related to the vehicle's operating environment. The vehicle and its associated controller use the detected information to navigate through the environment. The autonomous vehicle 101 can operate in a manual mode, in a fully autonomous mode, or in a partially autonomous mode.

[0030] In one embodiment, the autonomous vehicle 101 includes, but is not limited to, a perception and planning system 110, a vehicle control system 111, a wireless communication system 112, a user interface system 113, an infotainment system 114, and a sensor system 115. The autonomous vehicle 101 can also include certain common components included in ordinary vehicles, such as an engine, wheels, a steering wheel, a transmission, etc., which can be controlled by the vehicle control system 111 and / or the perception and planning system 110 using various communication signals and / or commands, such as acceleration signals or commands, deceleration signals or commands, steering signals or commands, braking signals or commands, etc.

[0031] Components 110-115 can be communicatively coupled to each other via an interconnect, a bus, a network, or a combination thereof. For example, components 110-115 can be communicatively coupled to each other via a controller area network (CAN) bus. The CAN bus is a vehicle bus standard designed to allow microcontrollers and devices to communicate with each other in applications without a host computer. It is a message-based protocol originally designed for multiplex electrical wiring within automobiles, but also used in many other contexts.

[0032] Referring now to Figure 2 In one embodiment, sensor system 115 includes, but is not limited to, one or more video cameras 211, a global positioning system (GPS) unit 212, an inertial measurement unit (IMU) 213, a radar unit 214, and a light detection and ranging (LIDAR) unit 215. GPS system 212 can include a transceiver operable to provide information about the location of the autonomous vehicle. IMU unit 213 can sense changes in position and orientation of the autonomous vehicle based on inertial acceleration. Radar unit 214 can represent a system that utilizes radio signals to sense objects within the local environment of the autonomous vehicle. In some embodiments, in addition to sensing objects, radar unit 214 can additionally sense the speed and / or direction of travel of the objects. LIDAR unit 215 can use lasers to sense objects in the environment in which the autonomous vehicle is located. In addition to other system components, LIDAR unit 215 can include one or more laser sources, a laser scanner, and one or more detectors. Video cameras 211 can include one or more devices to capture images of the environment surrounding the autonomous vehicle. Video cameras 211 can be still cameras and / or video cameras. The video cameras can be mechanically movable, for example, by mounting the video cameras on a rotating and / or tilting platform.

[0033] Sensor system 115 can also include other sensors, such as sonar sensors, infrared sensors, steering sensors, throttle sensors, brake sensors, and audio sensors (e.g., microphones). The audio sensors can be configured to capture sound from the environment surrounding the autonomous vehicle. The steering sensors can be configured to sense the steering angle of a steering wheel, a wheel of the vehicle, or a combination thereof. The throttle sensor and the brake sensor sense the throttle position and the brake position of the vehicle, respectively. In some cases, the throttle sensor and the brake sensor can be integrated as an integrated throttle / brake sensor.

[0034] In one implementation, the vehicle control system 111 includes, but is not limited to, a steering unit 201, a throttle unit 202 (also referred to as an acceleration unit), and a braking unit 203. The steering unit 201 is used to adjust the direction or heading of the vehicle. The throttle unit 202 is used to control the speed of the electric motor or engine, which in turn controls the speed and acceleration of the vehicle. The braking unit 203 slows down the vehicle by providing friction to slow down the wheels or tires of the vehicle. It should be noted that the components shown can be implemented in hardware, software, or a combination thereof. Figure 2

[0035] Referring back to Figure 1 The wireless communication system 112 allows for communication between the autonomous vehicle 101 and external systems such as devices, sensors, other vehicles, etc. For example, the wireless communication system 112 can wirelessly communicate directly with one or more devices, or wirelessly communicate via a communication network, such as communicating with the servers 103-104 through the network 102. The wireless communication system 112 can use any cellular communication network or wireless local area network (WLAN), e.g., using WiFi, to communicate with another component or system. The wireless communication system 112 can communicate directly with devices (e.g., a passenger's mobile device, a display device, a speaker within the vehicle 101), e.g., using an infrared link, Bluetooth, etc. The user interface system 113 can be part of the peripheral devices implemented within the vehicle 101, including, for example, a key, a touch screen display device, a microphone, and a speaker, etc.

[0036] Some or all of the functionality of the autonomous vehicle 101 can be controlled or managed by the perception and planning system 110, especially when operating in an autonomous mode. The perception and planning system 110 includes the necessary hardware (e.g., processors, memory, storage devices) and software (e.g., operating system, planning and routing programs) to receive information from the sensor system 115, the control system 111, the wireless communication system 112, and / or the user interface system 113, process the received information, plan a route or path from a starting point to a destination point, and then drive the vehicle 101 based on the planning and control information. Alternatively, the perception and planning system 110 can be integrated with the vehicle control system 111.

[0037] For example, a user who is a passenger can specify a starting location and a destination for a trip, e.g., via a user interface. The perception and planning system 110 obtains trip-related data. For example, the perception and planning system 110 can obtain location and routing information from a MPOI server, which can be part of the servers 103-104. A location server provides location services, and a MPOI server provides map services and POIs for certain locations. Alternatively, such location and MPOI information can be cached locally in a persistent storage of the perception and planning system 110.​

[0038] As the autonomous vehicle 101 moves along the route, the perception and planning system 110 can also obtain real-time traffic information from a traffic information system or server (TIS). It should be noted that the servers 103-104 can be operated by a third party entity. Alternatively, the functions of the servers 103-104 can be integrated with the perception and planning system 110. Based on the real-time traffic information, MPOI information and location information, as well as real-time local environment data (e.g., obstacles, objects, nearby vehicles) detected or sensed by the sensor system 115, the perception and planning system 110 can plan an optimal route and drive the vehicle 101 according to the planned route, e.g., via the control system 111, to safely and efficiently reach the designated destination.

[0039] The server 103 can be a data analytics system to perform data analytics services for various customers. In one implementation, the data analytics system 103 includes a data collector 121 and a machine learning engine 122. The data collector 121 collects driving statistics 123 from various vehicles (autonomous vehicles or regular vehicles driven by human drivers). The driving statistics 123 include information indicative of issued driving instructions (e.g., throttle, brake, steering instructions) as well as responses of the vehicle (e.g., speed, acceleration, deceleration, direction) captured by sensors of the vehicle at different points in time. The driving statistics 123 can also include information describing the driving environment at different points in time, e.g., route (including start location and destination location), MPOIs, road conditions, weather conditions, etc.

[0040] Based on the driving statistics 123, the machine learning engine 122 generates or trains a set of rules, algorithms and / or prediction models 124 for various purposes. In one implementation, the algorithms 124 can include algorithms for filtering candidate obstacles determined based on outputs of LIDAR devices. The algorithms 124 can then be uploaded to an ADV for real-time use during autonomous driving.

[0041] Figure 3A and Figure 3B is a block diagram illustrating an example of a perception and planning system for use with an autonomous vehicle, according to one implementation. The system 300 can be implemented as part of the autonomous vehicle 101 of Figure 1 , including but not limited to the perception and planning system 110, the control system 111 and the sensor system 115. Referring to Figures 3A-3B , the perception and planning system 110 includes but is not limited to a localization module 301, a perception module 302, a prediction module 303, a decision module 304, a planning module 305, a control module 306, a route scheduling module 307, an obstacle filtering module 403.

[0042] Some or all of the modules 301-307, 403 can be implemented in software, hardware, or a combination thereof. For example, the modules can be installed in a non-transitory storage device 352, loaded into memory 351, and executed by one or more processors (not shown). Note that some or all of the modules can be communicably coupled to or integrated with some or all of the modules of the vehicle control system 111 of FIG. 1. Figure 2 Some of the modules 301-307, 403 can be integrated together as an integrated module.

[0043] The localization module 301 determines the current location of the autonomous vehicle 300 (e.g., with the GPS unit 212) and manages any data related to a user's trip or route. The localization module 301, also referred to as a map and route module, manages any data related to a user's trip or route. A user can log in and specify a starting location and a destination of a trip, for example, via a user interface. The localization module 301 communicates with other components of the autonomous vehicle 300, such as the map and route information 311, to obtain trip-related data. For example, the localization module 301 can obtain location and route information from a location server and a map and POI (MPOI) server. The location server provides location services and the MPOI server provides map services and POIs for certain locations, which can be cached as part of the map and route information 311. The localization module 301 can also obtain real-time traffic information from a traffic information system or server as the autonomous vehicle 300 moves along a route.

[0044] Based on sensor data provided by the sensor system 115 and localization information obtained by the localization module 301, the perception module 302 determines a perception of the surrounding environment. The perception information can represent what an average driver would perceive around the vehicle the driver is driving. The perception can include, for example, lane configurations in the form of objects, traffic light signals, relative positions of other vehicles, pedestrians, buildings, crosswalks, or other traffic-related signs (e.g., stop signs, yield signs), etc. Lane configurations include information describing one or more lanes, such as, for example, a shape of a lane (e.g., straight or curved), a width of a lane, a number of lanes in a road, one-way or two-way lanes, merging or split lanes, exit lanes, etc.

[0045] The perception module 302 can include or function as a computer vision system to process and analyze images captured by one or more cameras to identify objects and / or features in the environment of the autonomous vehicle. Objects can include traffic signals, road boundaries, other vehicles, pedestrians, and / or obstacles, etc. The computer vision system can use object recognition algorithms, video tracking, and other computer vision techniques. In some embodiments, the computer vision system can map the environment, track objects, estimate the speed of objects, etc. The perception module 302 can also detect objects based on other sensor data provided by other sensors such as radar and / or LIDAR.

[0046] For each object, the prediction module 303 predicts how the object will behave in this situation. The prediction is performed based on the perception data that perceives the driving environment at the point in time considering a set of map / route information 311 and traffic rules 312. For example, if the object is a vehicle in the opposite direction and the current driving environment includes an intersection, the prediction module 303 will predict whether the vehicle is likely to move straight ahead or turn. If the perception data indicates that the intersection has no traffic lights, the prediction module 303 can predict that the vehicle can need to come to a complete stop before entering the intersection. If the perception data indicates that the vehicle is currently in a left-turn-only lane or a right-turn-only lane, the prediction module 303 can predict that the vehicle will be more likely to turn left or right, respectively.

[0047] For each object, the decision module 304 makes a decision about how to deal with the object. For example, for a particular object (e.g., another vehicle in the cross-traffic) and metadata describing the object (e.g., speed, direction, turn angle), the decision module 304 decides how to meet the object (e.g., pass, yield, stop, exceed). The decision module 304 can make such decisions according to a set of rules such as traffic rules or driving rules 312, which can be stored in the persistent storage 352.

[0048] The routing module 307 is configured to provide one or more routes or paths from a start point to a destination point. For a given trip from a start location to a destination location, such as a given trip received from a user, the routing module 307 obtains route and map information 311 and determines all possible routes or paths from the start location to the destination location. The routing module 307 can generate a reference line in the form of a topographic map that identifies each route from the start location to the destination location. The reference line refers to an ideal route or path that is free from any interference from other vehicles, obstacles, or traffic conditions. That is, if there are no other vehicles, pedestrians, or obstacles on the road, the ADV should follow the reference line exactly or closely. The topographic map is then provided to the decision module 304 and / or the planning module 305. The decision module 304 and / or the planning module 305 examine all possible routes to select and alter one of the best routes in accordance with other data provided by other modules, such as traffic conditions from the localization module 301, the driving environment perceived by the perception module 302, and traffic conditions predicted by the prediction module 303. Depending on the particular driving environment at a point in time, the actual path or route used to control the ADV can be close to or different from the reference line provided by the routing module 307.

[0049] Based on the decisions for each of the perceived objects, the planning module 305 uses the reference line provided by the routing module 307 as a basis to plan a path or route and driving parameters (e.g., distance, speed, and / or turn angle) for the autonomous vehicle. In other words, the decision module 304 decides what to do for a given object, and the planning module 305 determines how to do it. For example, for a given object, the decision module 304 can decide to pass the object, and the planning module 305 can determine whether to pass on the left or right side of the object. The planning and control data is generated by the planning module 305, including information describing how the vehicle 300 will move in the next movement cycle (e.g., the next route / path segment). For example, the planning and control data can indicate that the vehicle 300 is to move 10 meters at a speed of 30 miles per hour (mph), followed by a change to the right lane at a speed of 25 mph.

[0050] Based on the planning and control data, the control module 306 controls and drives the autonomous vehicle according to the route or path defined by the planning and control data by sending appropriate commands or signals to the vehicle control system 111. The planning and control data includes sufficient information to drive the vehicle from a first point to a second point along the path or route using appropriate vehicle settings or driving parameters (e.g., throttle, brake, steering commands) at different points in time.

[0051] In one embodiment, the planning phase is performed in a plurality of planning cycles (also referred to as driving cycles), e.g., in a cycle of every 100 milliseconds (ms) time interval. For each of the planning cycles or driving cycles, one or more control commands will be issued based on the planning and control data. That is, for every 100 ms, the planning module 305 plans a next route segment or path segment, e.g., including a target position and a time for the ADV to reach the target position. Alternatively, the planning module 305 can also specify a specific speed, direction, and / or steering angle, etc. In one embodiment, the planning module 305 plans a route segment or path segment for a next predetermined period of time, such as 5 seconds. For each planning cycle, the planning module 305 plans a target position for the current cycle (e.g., next 5 seconds) based on the target position planned in the previous cycle. The control module 306 then generates one or more control commands (e.g., throttle, brake, steering control commands) based on the planning and control data for the current cycle.

[0052] It is noted that the decision module 304 and the planning module 305 can be integrated as an integrated module. The decision module 304 / planning module 305 can include a navigation system or functionality of a navigation system to determine a driving path for the autonomous vehicle. For example, the navigation system can determine a series of speeds and heading directions for affecting movement of the autonomous vehicle along a path that substantially avoids perceived obstacles while advancing the autonomous vehicle along a roadway-based path leading to a final destination. The destination can be set according to user input via the user interface system 113. The navigation system can dynamically update the driving path while the autonomous vehicle is operating. The navigation system can combine data from a GPS system and one or more maps to determine the driving path for the autonomous vehicle.

[0053] The obstacle filtering module 403 filters the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise based at least in part on a feature associated with a point corresponding to each of the one or more candidate obstacles, as will be described in further detail below. The module 403 can be integrated with another of the modules 301-307, such as, for example, the perception module 302.

[0054] Referring to Figure 4FIG. 4, shows a block diagram 400 of various modules available, shown in accordance with one embodiment of the present disclosure. Based on the output of the LIDAR device / unit 215, a point cloud 401 comprising a plurality of points is generated. The point cloud 401 can represent a 3D shape of the characteristics of the surrounding environment. Each point can have its own set of X, Y, and Z coordinates and can be associated with an intensity measurement. The intensity measurement indicates the return intensity of the laser pulse that generated the point. Based on the point cloud at the perception module 302, one or more candidate obstacles 402 are determined. At the obstacle filtering module 403, the one or more candidate obstacles 402 are filtered to remove a first set of candidate obstacles of the one or more candidate obstacles 402 that correspond to noise based at least in part on a characteristic associated with the points corresponding to each of the one or more candidate obstacles 402. The noise can be caused by dust or other particles in the surrounding environment. In one embodiment, the first set of candidate obstacles that correspond to noise includes candidate obstacles that correspond to points that correspond to dust. In other words, since the noise can generate false positives of obstacles at the perception module 302, these false positives should be removed at the obstacle filtering module 403.

[0055] One or more identified obstacles 404 are determined that include the candidate obstacles that have not been removed at the obstacle filtering module 403. The operation of the autonomous vehicle is controlled by the control module 306 based on the identified obstacles 404.

[0056] In one embodiment, the characteristic associated with the points corresponding to each of the one or more candidate obstacles that the one or more candidate obstacles are filtered based on includes one or more of: a distribution of intensity measurements of the points corresponding to each of the one or more candidate obstacles, a spatial distribution of the points corresponding to each of the one or more candidate obstacles, or a combination thereof.

[0057] In one embodiment, filtering the one or more candidate obstacles includes removing each candidate obstacle that corresponds to a point in which a ratio of a number of points associated with low intensity measurements to a total number of points corresponding to the candidate obstacle is higher than a first threshold value. In one embodiment, the first threshold value can be approximately 0.9. The point is associated with low intensity measurements when the intensity measurement of the point is lower than an intensity threshold value. It should be understood that in other functionally equivalent embodiments, other ratios that capture the same numerical relationship can be used instead and the first threshold value can be adjusted accordingly with an inequality relationship. For example, a ratio of a number of points associated with low intensity measurements to a number of points not associated with low intensity measurements (e.g., points associated with high intensity measurements) can be used instead and the first threshold value can be adjusted accordingly. The actual ratio, inequality relationship, or first threshold value used does not limit the present disclosure.

[0058] Referring to Figure 5A and Figure 5B , a plot 500A, 500B showing a spatial distribution of LIDAR points corresponding to an obstacle is shown, in accordance with one embodiment. The LIDAR unit 215 emits a rotating laser pulse 502 to scan the environment. Based on the pulses reflected from the obstacle 501, a plurality of points 401 corresponding to the obstacle 501 can be determined. For a real obstacle such as the obstacle 501, the corresponding points 401 should collectively have a shape roughly similar to the capital letter L (or a rotated or flipped version of the capital letter L). This observation can be used to keep real obstacles and remove false positive candidate obstacles corresponding to points generated due to environmental noise such as dust or other particles, as will be described in further detail below. It should be further understood that for a real obstacle, the distribution of corresponding points at different heights varies relatively less, while for a group of points generated due to environmental noise, the distribution of points at different heights can vary significantly and correspond to a false positive candidate obstacle. This observation can also be used to remove false positive candidate obstacles, as will be described in further detail below. Figure 5B The projections 503A, 503B of the points 401 into two orthogonal dimensions (e.g., length side and width side) of the horizontal plane are shown, and respective standard deviations (б) of the projected points along the two-dimensional directions are determined. The standard deviations can be utilized to remove false positive candidate obstacles, as will be described in further detail below.

[0059] Referring back to Figure 4 , in one embodiment, at the obstacle filtering module 403, for each remaining candidate obstacle for which the identified obstacle type is a vehicle or the identified physical dimension is greater than a dimension threshold, the points corresponding to the candidate obstacle are projected into a first dimension and a second dimension of the horizontal plane. The first dimension and the second dimension can be, for example, the length side and the width side, respectively, and orthogonal to each other. A first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension are determined. The candidate obstacle is removed when the first standard deviation is higher than a first standard deviation threshold, or when the second standard deviation is higher than a second standard deviation threshold. In one embodiment, the candidate obstacle is removed when the first standard deviation is higher than the first standard deviation threshold and the second standard deviation is higher than the second standard deviation threshold.

[0060] In one embodiment, the first standard deviation threshold is equal to the second standard deviation threshold. In another embodiment, the first standard deviation threshold is different from the second standard deviation threshold. In one embodiment, either of the first standard deviation threshold and the second standard deviation threshold can be approximated to 0.6. Further, in functionally equivalent embodiments, other measures of dispersion other than standard deviation can be employed.

[0061] In one embodiment, for each remaining candidate obstacle whose identified obstacle type is a vehicle or whose identified physical dimension is greater than a dimension threshold, a region of a horizontal plane is projected to points corresponding to the candidate obstacle. The region of the horizontal plane is associated with a first number of squares. A second number of squares is determined within the first number of squares, each second square containing at least one projected point. The candidate obstacle is removed when a ratio of the second number of squares to the first number of squares is higher than a second threshold. In one embodiment, the second threshold can be approximately 0.6. It should be understood that in other functionally equivalent embodiments, other ratios that capture the same numerical relationship can be used instead, and the second threshold can be adjusted accordingly with an inequality relationship. For example, a ratio of a number of squares containing at least one projected point (i.e., the second number of squares) to a number of squares containing no projected points (i.e., the first number of squares minus the second number of squares) can be used instead, and the second threshold can be adjusted accordingly. The actual ratio, inequality relationship, or second threshold used does not limit the present disclosure.

[0062] In one embodiment, filtering the one or more candidate obstacles further includes removing each candidate obstacle corresponding to a point in which a ratio of a number of points associated with low height and a total number of points corresponding to the candidate obstacle is lower than a third threshold. In one embodiment, the third threshold can be approximately 0.5. The height of a point can correspond to a distance between the point and a ground plane or another predetermined reference plane. A point is associated with low height when the height of the point is lower than a height threshold. It should be understood that in other functionally equivalent embodiments, other ratios that capture the same numerical relationship can be used instead, and the third threshold can be adjusted accordingly with an inequality relationship. For example, a ratio of a number of points associated with low height and a number of points not associated with low height (e.g., points associated with high height) can be used instead, and the third threshold can be adjusted accordingly. The actual ratio, inequality relationship, or third threshold used does not limit the present disclosure.

[0063] In one embodiment, filtering the one or more candidate obstacles further includes removing each candidate obstacle that is not present in any perception result of a predetermined number of immediately preceding perception cycles (also referred to as frames). The number of immediately preceding perception cycles employed is a predetermined hyperparameter and can be determined empirically.

[0064] Various thresholds have been described in the foregoing. It should be understood that different values can be used for the various thresholds in different embodiments, and the actual thresholds do not limit the present disclosure.

[0065] Referring to Figure 6 FIG. 6 shows a flowchart of an exemplary method 600 for filtering candidate obstacles determined based on outputs of LIDAR devices in an autonomous vehicle, according to one embodiment.Figure 6 The process illustrated can be implemented in hardware, software, or a combination thereof. At box 610, a point cloud comprising multiple points is generated based on the output of the LiDAR device. At box 620, based on the point cloud, one or more candidate obstacles are identified. At box 630, the one or more candidate obstacles are filtered, at least in part, based on features associated with each corresponding point in the one or more candidate obstacles, to remove a first group of candidate obstacles corresponding to noise. At box 640, one or more identified obstacles, including candidate obstacles that have not yet been removed, are determined. At box 650, based on the identified obstacles, the operation of the autonomous vehicle is controlled.

[0066] Reference Figure 7 The diagram illustrates a flowchart of an exemplary method 700 for filtering candidate obstacles determined based on the output of a LIDAR device in an autonomous vehicle, according to one embodiment. Figure 7 The process shown can be implemented in hardware, software, or a combination thereof. Figure 6 Method 700 is executed at box 630. At box 710, each candidate obstacle corresponding to a point is removed if the ratio of the number of points associated with low intensity measurements to the total number of points corresponding to the candidate obstacle is higher than a first threshold. A point is associated with a low intensity measurement when its intensity measurement is lower than an intensity threshold. At box 720, for each remaining candidate obstacle whose identified obstacle type is a vehicle or whose identified physical size is greater than a size threshold, the point corresponding to that candidate obstacle is projected onto a first dimension and a second dimension of a horizontal plane. The first dimension and the second dimension are orthogonal to each other. At box 730, a first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension are determined. At box 740, a candidate obstacle is removed when the first standard deviation is higher than a first standard deviation threshold, or when the second standard deviation is higher than a second standard deviation threshold.

[0067] Therefore, according to embodiments of this disclosure, falsely reported candidate obstacles corresponding to LIDAR points that sequentially correspond to noise (such as noise caused by dust or other particles) can be removed. Removing falsely reported candidate obstacles reduces the possibility that the operation of the autonomous vehicle may be unduly interfered with by these falsely reported candidate obstacles.

[0068] It should be noted that some or all of the components illustrated and described above can be implemented in software, hardware or a combination thereof. For example, such components can be implemented as software installed and stored in a permanent storage device, which can be loaded in memory and executed by a processor (not shown) to implement the processes or operations throughout this application. Alternatively, such components can be implemented as executable code programmed or embedded in a special-purpose hardware, such as an integrated circuit (e.g., an application-specific integrated circuit or ASIC), a digital signal processor (DSP), or a field-programmable gate array (FPGA), which can be accessed via a corresponding driver and / or operating system from an application. Furthermore, such components can be implemented as specific hardware logic in a processor or processor core as part of an instruction set accessible by software components through one or more specific instructions.

[0069] Some portions of the foregoing detailed description have been presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. This description and representation have been in terms of operations on physical quantities, which are represented as physical manipulations by the machine.

[0070] It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the above discussion, it is appreciated that throughout the description, discussions utilizing terms such as those set forth in the claims below, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.

[0071] Embodiments of the disclosure also relate to an apparatus for performing the operations herein. Such computer program is stored in a non-transitory computer-readable medium. The machine-readable medium includes any mechanism for storing information in a form readable by a machine (e.g., a computer). For example, a machine-readable (e.g., computer-readable) medium includes a machine (e.g., a computer) readable storage medium (e.g., read only memory ("ROM"), random access memory ("RAM"), magnetic disk storage media, optical storage media, flash memory devices).

[0072] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces a means for implementing the functions / operations specified in the flowchart diagrams and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a standalone software package, or entirely on a remote machine or server.

[0073] The processes or methods depicted in the preceding figures can be performed by processing logic that comprises hardware (e.g., circuitry, dedicated logic, etc.), software (e.g., embodied on a non-transitory computer readable medium), or a combination of both. Although the processes or methods are described above in terms of some sequential operations, it should be appreciated that some of the operations described can be performed in different order. Moreover, some of the operations can be performed in parallel rather than sequentially.

[0074] The implementations of the disclosure have not been described with reference to any particular programming language. It will be appreciated that a variety of programming languages can be used to implement the teachings of the implementations of the disclosure as described herein.

[0075] In the foregoing specification, embodiments of the disclosure have been described with reference to specific exemplary embodiments thereof. It will be evident that various modifications can be made to the disclosure without departing from the broader spirit and scope of the disclosure as set forth in the appended claims. The specification and drawings are, accordingly, to be regarded in an illustrative sense rather than a restrictive sense.

Claims

1. A computer-implemented method comprising: generating, based on an output of a light detection and ranging device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering, based at least in part on a characteristic associated with points corresponding to each of the one or more candidate obstacles, the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise, including, for each remaining candidate obstacle for which an identified obstacle type is a vehicle or an identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to a first dimension and a second dimension of a horizontal plane, the first dimension and the second dimension being orthogonal to each other, determining a first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension, and removing the candidate obstacle when the first standard deviation is higher than a first standard deviation threshold or when the second standard deviation is higher than a second standard deviation threshold, the characteristic including one or more of a distribution of intensity measurements of the points corresponding to each of the one or more candidate obstacles, a spatial distribution of the points corresponding to each of the one or more candidate obstacles, or a combination thereof; determining one or more identified obstacles comprising the candidate obstacles that have not been removed; controlling, based on the identified obstacles, an operation of an autonomous vehicle.

2. The method of claim 1, wherein, the first set of candidate obstacles corresponding to noise includes candidate obstacles corresponding to points corresponding to dust.

3. The method of claim 1, wherein, filtering the one or more candidate obstacles includes: removing each candidate obstacle corresponding to points for which a ratio of a number of points associated with low intensity measurements to a total number of points corresponding to the candidate obstacle is higher than a first threshold, wherein a point is associated with low intensity measurements when an intensity measurement of the point is lower than an intensity threshold.

4. The method of claim 1, wherein, the candidate obstacle is removed when the first standard deviation is higher than the first standard deviation threshold and the second standard deviation is higher than the second standard deviation threshold.

5. The method of claim 1, wherein, the first standard deviation threshold is equal to the second standard deviation threshold.

6. The method of claim 1, wherein, the first standard deviation threshold is different from the second standard deviation threshold.

7. The method of claim 1, wherein, filtering the one or more candidate obstacles further includes, for each remaining candidate obstacle for which an identified obstacle type is a vehicle or an identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to an area of the horizontal plane, the area of the horizontal plane being associated with a first square number of grids; determining, within the first square number, a second square number, each of the second square number containing at least one projected point; and the candidate obstacle is removed when a ratio of the second square number to the first square number is higher than a second threshold.

8. The method of claim 7, wherein, filtering the one or more candidate obstacles further includes: removing each candidate obstacle that corresponds to points in which a ratio of a number of points associated with low height to a total number of points corresponding to the candidate obstacle is below a third threshold, wherein the points are associated with low height when a height of the points is below a height threshold.

9. The method of claim 8, wherein, filtering the one or more candidate obstacles further comprises: removing each candidate obstacle that corresponds to points in which a ratio of a number of points associated with low height to a total number of points corresponding to the candidate obstacle is below a third threshold, wherein the points are associated with low height when a height of the points is below a height threshold.

10. A non-transitory machine-readable medium having stored therein instructions which, when executed by a processor, cause the processor to perform operations comprising: generating, based on an output of a light detection and ranging device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering the one or more candidate obstacles to remove a first set of candidate obstacles of the one or more candidate obstacles corresponding to noise based at least in part on a characteristic associated with points corresponding to each of the one or more candidate obstacles, including, for each remaining candidate obstacle for which an identified obstacle type is a vehicle or an identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to a first dimension and a second dimension of a horizontal plane, the first dimension and the second dimension being orthogonal to each other, determining a first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension, and removing the candidate obstacle when the first standard deviation is above a first standard deviation threshold or when the second standard deviation is above a second standard deviation threshold, the characteristic including one or more of a distribution of intensity measurements of the points corresponding to each of the one or more candidate obstacles, a spatial distribution of the points corresponding to each of the one or more candidate obstacles, or a combination thereof; determining one or more identified obstacles including the candidate obstacles that have not been removed; controlling an operation of an autonomous vehicle based on the identified obstacles.

11. The non-transitory machine-readable medium of claim 10, wherein, the first set of candidate obstacles corresponding to noise includes candidate obstacles corresponding to points corresponding to dust.

12. The non-transitory machine-readable medium of claim 10, wherein, filtering the one or more candidate obstacles further comprises: removing each candidate obstacle that corresponds to points in which a ratio of a number of points associated with low intensity measurement to a total number of points corresponding to the candidate obstacle is above a first threshold, wherein the points are associated with low intensity measurement when an intensity measurement of the points is below an intensity threshold.

13. The non-transitory machine-readable medium of claim 10, wherein, the candidate obstacle is removed when the first standard deviation is above the first standard deviation threshold and the second standard deviation is above the second standard deviation threshold.

14. The non-transitory machine-readable medium of claim 10, wherein, the first standard deviation threshold is equal to the second standard deviation threshold.

15. The non-transitory machine-readable medium of claim 10, wherein, the first standard deviation threshold is different from the second standard deviation threshold.

16. The non-transitory machine-readable medium of claim 10, wherein, filtering the one or more candidate obstacles further comprises, for each remaining candidate obstacle for which an identified obstacle type is a vehicle or an identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to a region of the horizontal plane, the region of the horizontal plane being associated with a first number of squares; determining a second number of squares within the first number of squares, each of the second number of squares containing at least one projected point; and removing the candidate obstacle when a ratio of the second number of squares to the first number of squares is higher than a second threshold.

17. The non-transitory machine-readable medium of claim 16, wherein, filtering the one or more candidate obstacles further comprises, removing each candidate obstacle corresponding to points in which a ratio of a number of points associated with low height to a total number of points corresponding to the candidate obstacle is lower than a third threshold, wherein a point is associated with low height when a height of the point is lower than a height threshold.

18. The non-transitory machine-readable medium of claim 17, wherein, filtering the one or more candidate obstacles further comprises: removing each candidate obstacle that is not present in perception results of any of a predetermined number of immediately preceding perception cycles.

19. A data processing system comprising: a processor; and a memory coupled to the processor to store instructions that, when executed by the processor, cause the processor to perform operations comprising: generating, based on an output of a light detection and ranging device, a point cloud comprising a plurality of points; determining, based on the point cloud, one or more candidate obstacles; filtering, based at least in part on a characteristic associated with points corresponding to each of the one or more candidate obstacles, the one or more candidate obstacles to remove a first group of candidate obstacles of the one or more candidate obstacles corresponding to noise, including, for each remaining candidate obstacle for which an identified obstacle type is a vehicle or an identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to a first dimension and a second dimension of a horizontal plane, the first dimension and the second dimension being orthogonal to each other; determining a first standard deviation of the projected points along the first dimension and a second standard deviation of the projected points along the second dimension; and removing the candidate obstacle when the first standard deviation is higher than a first standard deviation threshold or when the second standard deviation is higher than a second standard deviation threshold; the characteristic including one or more of a distribution of intensity measurements of the points corresponding to each of the one or more candidate obstacles, a spatial distribution of the points corresponding to each of the one or more candidate obstacles, or a combination thereof; determining one or more identified obstacles comprising the candidate obstacles that have not been removed; controlling, based on the identified obstacles, an operation of an autonomous vehicle.

20. The system of claim 19, wherein, the first group of candidate obstacles corresponding to noise includes candidate obstacles corresponding to points corresponding to dust.

21. The system of claim 19, wherein, filtering the one or more candidate obstacles includes: removing each candidate obstacle corresponding to points in which a ratio of a number of points associated with low intensity measurements to a total number of points corresponding to the candidate obstacle is higher than a first threshold, wherein a point is associated with low intensity measurements when an intensity measurement of the point is lower than an intensity threshold.

22. The system of claim 19, wherein, removing the candidate obstacle when the first standard deviation is higher than the first standard deviation threshold and the second standard deviation is higher than the second standard deviation threshold.

23. The system of claim 19, wherein, the first standard deviation threshold is equal to the second standard deviation threshold.

24. The system of claim 19, wherein, the first standard deviation threshold is different from the second standard deviation threshold.

25. The system of claim 19, wherein, filtering the one or more candidate obstacles further comprises, for each remaining candidate obstacle for which the identified obstacle type is a vehicle or the identified physical dimension is greater than a dimension threshold, projecting points corresponding to the candidate obstacle to a region of the horizontal plane, the region of the horizontal plane being associated with a first square number of grids; determining a second square number within the first square number, each of the second square number containing at least one projected point; and removing the candidate obstacle when a ratio of the second square number and the first square number is higher than a second threshold.

26. The system of claim 25, wherein, filtering the one or more candidate obstacles further comprises: removing each candidate obstacle for which a ratio of a number of points associated with low height and a total number of points corresponding to the candidate obstacle is lower than a third threshold, wherein a point is associated with low height when a height of the point is lower than a height threshold.

27. The system of claim 26, wherein, filtering the one or more candidate obstacles further comprises: removing each candidate obstacle for which a ratio of a number of points associated with low height and a total number of points corresponding to the candidate obstacle is lower than a third threshold, wherein a point is associated with low height when a height of the point is lower than a height threshold. filtering the one or more candidate obstacles further comprises: removing each candidate obstacle for which a ratio of a number of points associated with low height and a total number of points corresponding to the candidate obstacle is lower than a third threshold, wherein a point is associated with low height when a height of the point is lower than a height threshold.

28. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-9.

Images