Lane change planning and control in autonomous machine applications

Abstract

In various examples, sensor data can be collected using one or more sensors of the ego vehicle to generate a representation of the ego vehicle's surroundings. The representation can include lanes of a roadway and object locations within the lanes. The representation of the environment can be provided as input to a longitudinal velocity profile identifier, which can project a plurality of longitudinal velocity profile candidates onto a target lane. Each of the plurality of longitudinal velocity profile candidates can be evaluated one or more times based on one or more sets of criteria. Using scores from the evaluations, a target gap and a particular longitudinal velocity profile can be selected from the longitudinal velocity profile candidates. Once the longitudinal velocity profile for the target gap is determined, the system can perform a lane change maneuver in accordance with the longitudinal velocity profile.

Description

BACKGROUND

[0001] Vehicles capable of operating with little to no human input often need to perform autonomous lane changes. The ability to perform autonomous lane changes is a key feature of all levels of autonomous driving. During a lane change maneuver, an autonomous vehicle (“ego-vehicle”) must be able to accomplish two important planning and control tasks. The first important task is to create a speed plan that enables it to adjust its speed in order to safely and timely perform the lane change maneuver. Without a longitudinal speed adaptation plan, the ego-vehicle can not have enough information to perform operations such as how to slow down and fill the gap behind a leading vehicle in the target lane, or equivalently, how to accelerate past a trailing vehicle in the target lane and find a space to change lanes. The second important task is to create a path plan that enables the ego-vehicle to move from the ego-lane to the target lane in a way that utilizes and adheres to the speed plan. While the ego-vehicle can be able to perceive its environment, the vehicle can not be able to determine how to adjust its speed and path to enable the vehicle to safely perform the lane change operation in a timely and efficient manner.

[0002] For example, some conventional systems use deep learning, such as one or more deep neural networks (DNNs), to perform a lane change, e.g., to compute an output trajectory for the vehicle to follow through the lane change. However, these DNNs often are not able to generate a lane change trajectory that results in a comfortable (e.g., below jerk or acceleration limits) lane change maneuver, and can not take into account surrounding vehicles or identify actual positions in the target lane that the ego-vehicle can maneuver into (e.g., the identified positions can be occupied at the time the maneuver is performed). For example, the DNNs can not take into account the speed or acceleration of a leading or trailing vehicle, so even though the lane change maneuver output by the DNN can result in a lane change maneuver, the maneuver can be aborted during execution as the trailing vehicle rapidly approaches the ego-vehicle or the leading vehicle slows down into the ego-vehicle’s path. As a result, the lane change plans of these conventional systems can not result in efficient, safe, and / or comfortable lane change maneuvers due to not adequately taking into account the trajectories of other vehicles or objects in the environment. SUMMARY

[0003] Embodiments of the present disclosure relate to lane change planning and control in autonomous machine applications. Systems and methods are disclosed that can identify a longitudinal speed profile of a target gap by evaluating a plurality of longitudinal speed profile candidates in view of one or more criteria.

[0004] In contrast to traditional systems such as those described above, the present system identifies longitudinal velocity profiles for performing a lane change maneuver. For example, as the ego vehicle travels along a roadway, the system of the ego vehicle can receive an input signal - e.g., from a user, a navigation application, and / or a lane change command connected to the system of the ego vehicle - to perform a lane change operation. Based on the input signal, sensor data can be collected using one or more sensors of the ego vehicle to generate an environmental model or representation, which can include the roadway lanes and the locations of objects within the lanes. The representation of the environment can also be used to predict a future location or trajectory of each object. For example, by determining a velocity corresponding to each object, a future location of each object at each time t over a period of time can be determined to effectively create a future prediction model of the environment.

[0005] The representation of the environment can be provided as input to a longitudinal velocity profile identifier for selecting a longitudinal velocity profile, which can include acceleration and / or deceleration commands to help complete a lane change from the ego lane to a target lane available for longitudinal space of the ego vehicle. For example, upon receiving a lane change command, the longitudinal velocity profile identifier can identify one or more candidate lane change gaps in the target lane and project a plurality of longitudinal velocity profiles onto the target lane. Each of the plurality of longitudinal velocity profiles can then be evaluated based on a set of criteria. A cumulative score for each longitudinal velocity profile for each lane change gap can then be used to select a target lane change gap for a current lane change maneuver. The selected target lane change gap can then be evaluated to determine a particular longitudinal velocity profile from the longitudinal velocity profile candidates. For example, in some embodiments, the longitudinal velocity profile candidates can then be re-evaluated for the target lane change gap to determine a longitudinal velocity profile from the longitudinal velocity profile candidates for performing a lane change maneuver into the target lane change gap in the target lane. Once the longitudinal velocity profile and the target lane change gap are determined, the system can perform the lane change maneuver according to the longitudinal velocity profile.

[0006] Accordingly, the systems and methods of the present disclosure can evaluate longitudinal velocity profiles in order to control the ego vehicle towards a position in the ego lane that increases the probability of a successful, safe, and comfortable lane change maneuver into a target lane. For example, by decelerating or accelerating the ego vehicle to align the ego vehicle with a lane change gap, the start of a lateral lane change maneuver can be performed from a position in the ego lane that is more likely to result in a successful maneuver. BRIEF DESCRIPTION OF DRAWINGS

[0007] The present systems and methods for lane change planning and control in autonomous machine applications are described in detail below with reference to the attached drawing figures, wherein:

[0008] Figure 1is a data flow diagram of a system for performing a lane change maneuver according to some embodiments of the present disclosure;

[0009] Figure 2 depicts an example flow diagram for selecting a longitudinal speed profile according to some embodiments of the present disclosure;

[0010] Figure 3A depicts a top view of an example lane change scenario according to embodiments of the present disclosure;

[0011] Figure 3B depicts a one-dimensional (ID) lane change scenario according to embodiments of the present disclosure;

[0012] Figure 4 is a flow diagram illustrating a method for identifying a longitudinal speed profile according to some embodiments of the present disclosure;

[0013] Figure 5A is an illustration of an example autonomous vehicle according to some embodiments of the present disclosure;

[0014] Figure 5B is an example of a camera position and field of view of an example autonomous vehicle according to some embodiments of the present disclosure; Figure 5A

[0015] Figure 5C is a block diagram of an example system architecture of an example autonomous vehicle according to some embodiments of the present disclosure; Figure 5A

[0016] Figure 5D is a system diagram of communication between a cloud-based server and an example autonomous vehicle according to some embodiments of the present disclosure; Figure 5A

[0017] Figure 6 is a block diagram of an example computing device suitable for implementing some embodiments of the present disclosure; and

[0018] Figure 7 is a block diagram of an example data center suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0019] Systems and methods related to lane change planning and control in autonomous machine applications are disclosed. Although the present disclosure can be directed to example autonomous vehicle 500 (also referred to herein as "vehicle 500" or "my vehicle 500," examples of which are described herein with respect to Figures 5A-5D ​​​Described, this is not limiting. For example, the systems and methods described herein can be used by, but are not limited to, non-autonomous vehicles, semi-autonomous vehicles (e.g., in one or more adaptive driver assistance systems (ADAS)), vehicles coupled to a trailer, manned and unmanned robots or robotic platforms, warehouse vehicles, off-road vehicles, flying vessels, watercraft, shuttle buses, emergency vehicles, motorcycles, electric or motorized bicycles, airplanes, engineering vehicles, underwater vehicles, drones, and / or other vehicle types. Furthermore, while the present disclosure can be described with respect to lane change maneuvers for vehicle applications, this is not intended to be limiting and the systems and methods described herein can be used to enhance realty, virtual reality, robotics, robotic assistance platforms, autonomous or semi-autonomous machine applications, and / or any other technology space that can use longitudinal and / or lateral maneuver trajectories or path planning.

[0020] In some embodiments, as the ego vehicle travels along a roadway, systems of the ego vehicle can receive input signals - e.g., from a user, a navigation application, and / or a lane change command connected to the systems of the ego vehicle - to perform a lane change operation. Based on the input signals, sensor data can be collected using one or more sensors of the ego vehicle and provided to an object detector to detect and / or track objects (e.g., vehicles, pedestrians, bicyclists, motorcyclists, etc.) in the environment around the ego vehicle. Furthermore, a lane identifier can generate a lane graph or other lane representation using perception information, map information (e.g., high definition (HD) map information), and / or other sources of information. Outputs from the object detector and the lane identifier can be provided to an environment modeler, which can generate models or representations of the roadway lanes and the locations of the objects within the lanes. For example, these models or representations can reflect which objects are assigned to which lanes around the ego vehicle, as well as the speed at which the objects are moving relative to the ego vehicle and / or the predicted or estimated future trajectories of the objects.

[0021] By computing a bounding shape (e.g., a box, square, rectangle, triangle, circle, polygon, etc.) for each object and cropping a portion of the bounding shape, the object can be assigned to a lane shape in the lane graph (or other lane representation) so that the bounding shape more accurately reflects the portion of the object that is closest to the driving surface. For example, the computed drivable free space information can be used to crop the bounding shape so that only the lower portion of the bounding shape is retained. In other examples, a percentage or other amount of the bounding shape can be cropped, such as the upper 70%, 80%, 95%, etc. of the bounding shape. The resulting shape after cropping can be referred to as an object fence.

[0022] The object fence locations can be compared to the lane graph to determine the location of each object in the lane graph - e.g., similar to object-in-path analysis (OIPA). The objects can then be associated with the lane graph at the determined locations to generate a representation of the ego vehicle’s environment. In some embodiments, the representation of the environment can include only the ego lane and one or more adjacent lanes to the ego lane, or can include only target lane candidates for the lane change maneuver. Thus, the processing requirements can be reduced, as the system can only consider potential target lanes in the lane change maneuver plan. However, in other embodiments, any number of lanes can be included in the representation of the environment.

[0023] In some embodiments, the representation of the environment can include a top-down projection of the lane graph (or other representation type) and object fence information corresponding to detected objects, represented as beads. This information can also be used to predict a future location or trajectory of each object. For example, by determining a velocity corresponding to each bead, a future location of each bead at each time t in a period of time can be determined, effectively creating a future prediction model of the lane change environment.

[0024] The prediction can be modeled as a one-dimensional (ID) longitudinal projection of the lane graph (e.g., a two-dimensional (2D) environment model). For example, the system can compress the ego lane and the target lane into a one-dimensional lane graph by projecting the longitudinal motion of the ego vehicle and ego-leading objects (e.g., objects leading the ego vehicle in the ego lane, if present) from the ego lane to the target lane. The beads representing object fence information can be projected onto the one-dimensional lane graph, which can allow the system to determine gaps in the target lane, e.g., by determining one or more spaces between the beads on the lane graph. In some embodiments, multiple types of beads can be included in the lane graph for the same physical object at a given location by using multiple sensor types for object detection. For example, one bead can correspond to one object from one sensor type (e.g., a camera sensor), while another bead can correspond to the same object, but can be generated from data from another sensor type (e.g., a RADAR sensor, a LiDAR sensor, etc.). Depending on the embodiment, two or more beads can be used individually, or can be combined - e.g., using clustering - prior to use.

[0025] In some embodiments, the representation of the environment can be provided as input to a longitudinal speed profile identifier for selecting a longitudinal speed profile. The longitudinal speed profile can include acceleration and / or deceleration commands to help accomplish a lane change from a self-lane to a target lane in longitudinal space available to the ego vehicle. When a lane change command is received, the longitudinal speed profile identifier can use the representation of the environment to identify one or more candidate lane change gaps (or referred to as “gaps” herein) in the target lane, which can then be evaluated for each candidate lane change gap. The evaluation of a candidate gap can include, for each candidate gap, projecting a plurality (e.g., 100, 250, 700, 5000, etc.) of longitudinal speed profile candidates to the target lane, which can be represented by a ID lane graph. In embodiments, each longitudinal speed profile candidate can be represented as an “S” curve, where one portion of the speed profile can correspond to acceleration and another portion can correspond to deceleration. In some embodiments, the deceleration portion can always be the first portion and the acceleration portion can always be the second portion (e.g., because if the ego vehicle is traveling at a speed limit, the first speed adaptation can not accelerate without violating speed laws). However, there can be no acceleration portion or deceleration portion depending on the longitudinal speed profile. Further, in some embodiments, the acceleration portion can precede the deceleration portion, or the profile can include any number of acceleration portions and / or any order of deceleration portions.

[0026] Each longitudinal speed profile candidate can be evaluated (e.g., scored) for each gap based on a set of criteria (e.g., speed adaptation duration, acceleration limit, deceleration limit, speed vector field, fast moving limit, etc.) to generate a cumulative score for each gap. For example, with 5000 longitudinal speed profile candidates, the same 5000 candidates can be evaluated for each gap. The cumulative score for each gap can then be used to select a target gap for a current lane change maneuver. For example, the candidate gap with the best (e.g., highest, lowest, etc., depending on the scoring system) cumulative score can be selected as the target gap.

[0027] The selected target gap can then be evaluated to determine a particular longitudinal velocity profile from the longitudinal velocity profile candidates in accordance with the longitudinal velocity profile candidates. For example, in embodiments, the longitudinal velocity profile candidates can then be reevaluated for the target gap to determine a longitudinal velocity profile from the longitudinal velocity profile candidates for performing a lane change maneuver into the target gap of the target lane. In some embodiments, each longitudinal lane change candidate can be reevaluated for the target gap, or in other embodiments, a subset of the longitudinal lane change candidates can be evaluated in consideration of the target gap. For example, in cases where a subset is used, the subset can have been filtered from a larger group during the gap evaluation process. As such, longitudinal velocity profile candidates evaluated for the target gap that do not violate the velocity vector field (e.g., do not result in the ego vehicle colliding with the ego lead object) and / or satisfy other criteria during the gap evaluation process that results in the ego vehicle being in the target gap (e.g., not outside of the target gap, not colliding with an object, etc.) can be used as the subset of longitudinal velocity profile candidates. In either example, the longitudinal velocity profile candidates evaluated for the target gap can be evaluated against another set (e.g., different, same, etc.) of criteria (which can include at least some overlap criteria with the gap evaluation) that can include a speed adaptation duration, an acceleration limit, a deceleration limit, a fast moving limit, a stop distance overlap, a courtesy distance, a back-off gap count, or a speed loss amount to determine a velocity profile with the best score. Each longitudinal velocity profile candidate - or the subset thereof - can be scored based on the evaluation taking into account some or all of these criteria, and the longitudinal velocity profile candidate with the best (e.g., highest, lowest, etc., depending on the embodiment) score can be selected as the longitudinal velocity profile.

[0028] Once the longitudinal velocity profile and the target gap are determined, the system can perform the lane change maneuver in accordance with the longitudinal velocity profile - e.g., first decelerating then accelerating, etc. In addition to the longitudinal velocity profile, the system can evaluate a plurality of lateral profile candidates. For example, in each frame or iteration, the longitudinal velocity profile identifier and the lateral profile identifier can communicate with each other (e.g., the identifiers can be bidirectionally communicatively coupled and exchange intermediate information) in order to identify a lateral profile that matches the longitudinal velocity profile - e.g., to allow for a lane change maneuver in accordance with the longitudinal velocity profile that is comfortable, safe, etc.

[0029] In some embodiments, the longitudinal velocity profile, or at least a portion thereof, can be executed first in order to align the ego vehicle with the target gap while remaining in the ego lane. In such examples, once aligned, the identified lateral profile can be executed in conjunction with another longitudinal velocity profile (or other portion of the initial longitudinal velocity profile) to complete the lane change maneuver into the target lane. Similar to the longitudinal velocity profile candidates, multiple lateral profile candidates can be evaluated according to one or more criteria - e.g., lateral deviation from the target lane, lateral acceleration limit, convergence speed to the center of the target lane, potential lateral collision, etc. - to find the lateral profile with the highest score.

[0030] The selected longitudinal velocity profile and lateral profile at each frame can be used to execute the lane change maneuver from the ego lane to the target gap of the target lane. As a result, the lane change maneuver can take into account past, current, and future trajectories of objects surrounding the ego vehicle, such that a safe and effective lane change maneuver can be executed. Moreover, by determining and executing a longitudinal velocity profile that aligns the ego vehicle with the target lane change gap while still in the ego lane, the success rate of the lane change maneuver can be improved.

[0031] Reference Figure 1 , Figure 1 is a data flow diagram of a system 100 for performing a lane change maneuver according to some embodiments of the present disclosure. It should be understood that this and other arrangements described herein are set forth only as examples. Other arrangements and elements (e.g., machines, interfaces, functions, orders, groupings of functions, etc.) can be used in addition to or instead of those shown, and some elements can be wholly omitted or consolidated. In addition, many of the elements described herein are functional entities that can be implemented as discrete or distributed components or in conjunction with other components, and in any suitable combination and location. Various functions described herein as being performed by entities can be performed by hardware, firmware, and / or software. For instance, various functions can be performed by a processor executing instructions stored in memory.

[0032] Figure 1The data flow diagram of FIG. 1 includes sensor data 110, object detector 112, lane identifier 114, environment modeler 116, longitudinal velocity profile identifier 118, lateral path identifier 120, safety checker 122, lane change trajectory determiner 124, and vehicle control 126. In operation, system 100 can include generating and / or receiving sensor data 110 from one or more sensors, such as sensors of vehicle 500. Sensor data 110 can be used by object detector 112 and / or lane identifier 114 to generate data for use by environment modeler 116, such as to generate a representation of the ego vehicle’s environment, including one or more lanes, current positions of objects (e.g., other vehicles) in the lanes, and / or future positions / trajectories of the objects. The modeled environment, such as the representation of the environment, can be used by longitudinal velocity profile identifier 118 and / or lateral path identifier 120 to generate data that can be used by lane change trajectory determiner 124 to determine a trajectory for ego vehicle 500 through a lane change maneuver from the ego lane to a target lane of a lane change.

[0033] In some embodiments, sensor data 110 can include, without limitation, data from vehicle 500 (and / or other vehicles, machines, or objects, such as robotic devices, watercraft, aircraft, trains, in some examples, construction equipment, VR systems, AR systems, etc.). For non-limiting examples, such as where the sensors generating sensor data 110 are disposed on or otherwise associated with a vehicle, sensor data 110 can include, without limitation, data generated by global navigation satellite systems (GNSS), sensor 558 (e.g., a global positioning system sensor), RADAR sensor 560, ultrasonic sensor 562, LIDAR sensor 564, inertial measurement unit (IMU) sensor 566 (e.g., an accelerometer, a gyroscope, a magnetic compass, a magnetometer, etc.), microphone 596, stereo camera 568, wide-angle camera 570 (e.g., a fisheye camera), infrared camera 572, surround camera 574 (e.g., a 360-degree camera), long and / or medium range camera 598, speed sensor 544 (e.g., to measure a speed of vehicle 500), and / or other sensor types.

[0034] In some examples, sensor data 110 can include sensor data generated by one or more forward-facing sensors, side-viewing sensors, and / or rear-viewing sensors. This sensor data 110 can be used by object detector 112, lane identifier 114, environment modeler 116, and / or one or more other components of system 100 to help determine a lane change trajectory for vehicle 500. In embodiments, any number of sensors can be used to incorporate multiple fields of view (e.g., Figure 5Ba sensing field (e.g., of a LIDAR sensor 564, RADAR sensor 560, etc.) and / or a field of view (e.g., of a long-range camera 598, a forward-facing stereo camera 568, and / or a forward-facing wide-view camera 570).

[0035] The object detector 112 can use the sensor data 110 to detect and / or track objects (e.g., vehicles, pedestrians, bicyclists, motorcyclists, trucks, etc.) in the ego vehicle’s environment. For example, one or more computer vision algorithms, machine learning models, and / or deep neural networks (DNNs) can be implemented by the object detector 112 to detect objects and / or track objects over time. In some embodiments, the object detector 112 can be used to estimate or predict future locations or trajectories of objects in the environment.

[0036] The lane identifier 114 can use the sensor data 110 and / or map information, such as from a Global Navigation Satellite System (GNSS) map, a high-definition (HD) map, another map type that is capable of providing a near 1 : 1 scale of a real-world environment, and / or other map types, to determine a number of lanes, a type of lane, a location of a lane, and / or otherwise identify lanes or other delineated driving regions in the environment. For example, in cases where map data is used, the sensor data 110 and map data can be used to localize the ego vehicle 500 within the map, and lanes can be determined from the map after localization. In some embodiments, the lane identifier 114 can generate a lane graph representing lanes in the environment.

[0037] The environment modeler 116 can generate models or representations of lanes of a road and object locations within the lanes based on inputs from the object detector 112 and / or the lane identifier 114. For example, these models or representations can reflect which objects are assigned to which lanes around the ego vehicle 500, as well as speeds at which the objects are moving relative to the ego vehicle 500 and / or predicted or estimated future trajectories of the objects. In addition to output from the object detector 112 to assign objects to lanes in a lane graph (or other lane representation), the environment modeler 116 can generate a lane graph (and / or use a lane graph generated by the lane identifier 114) to generate a representation of the environment. In some embodiments, objects can be assigned to respective locations in lanes in the lane graph by computing a bounding shape (e.g., a box, square, rectangle, triangle, circle, polygon, etc.) for each detected object and cropping portions of the bounding shape so that the bounding shape more accurately reflects portions of the object that are closest to the driving surface.

[0038] In some embodiments, the free space filter of the environment modeler 116 can be used to filter out image portions corresponding to representations of non-drivable or non-navigable space or road lanes within the model and / or to filter out portions or subsets of points in the point cloud that correspond to non-drivable or non-navigable space. For example, sidewalks, buildings, other vehicles, pedestrians, bicyclists, animals, trees, and / or other portions of the environment corresponding to the model or representation of road lanes can be filtered out or ignored, and drivable surfaces - e.g., roads, lanes, parking lots, driveways, etc. - can be left for further analysis. In some examples, the computed drivable free space information can be used to crop the boundary shape such that only a portion of the boundary shape corresponding to drivable free space is retained. Additionally or alternatively, a certain percentage or other amount of the boundary shape can be cropped - e.g., the top 70%, 80%, 95%, etc. of the boundary shape. The resulting shape after cropping can be referred to as an object fence.

[0039] The environment modeler 116 can compare the object fence locations to the lane graph to determine the location of each object in the lane graph - e.g., similar to object-in-path analysis (OIPA). The environment modeler 116 can then associate the objects with the lane graph at the determined locations to generate a representation of the environment of the ego vehicle 500. In some embodiments, the representation or model of the environment generated by the environment modeler 116 can be limited to the ego lane and one or more lanes adjacent to the ego lane, or can be limited to only the target lane for a lane change maneuver. However, in other embodiments, the representation of the environment can include any number of lanes at any location relative to the ego vehicle 500.

[0040] In some embodiments, the environment modeler 116 can generate a representation of the environment that includes a top-down projection of the lane graph (or other representation type) and object fence information corresponding to detected objects, which can be represented as beads. The environment modeler 116 can further use this information to predict future locations or trajectories of each object on the road relative to the ego vehicle 500.

[0041] In some embodiments, the environment modeler 116 can model the prediction of the future position or trajectory of each object on the road as a one-dimensional (ID) longitudinal projection of a lane graph (e.g., a two-dimensional (2D) environment model). For example, the environment modeler 116 can compress the ego lane and the target lane into a one-dimensional lane graph by projecting the longitudinal motion of the ego vehicle 500 and ego-leading objects (e.g., objects leading the ego vehicle in the ego lane, if present) from the ego lane to the target lane. The environment modeler 116 can project beads representing object fence information onto the ID lane graph, which can allow the environment modeler 116 to determine a lane change gap in the target lane, e.g., by determining one or more spaces between the beads on the lane graph. In some embodiments, multiple types of beads can be included in the lane graph for the same physical object at a given location by using multiple sensor types for object detection. For example, one bead can correspond to one object from one sensor type (e.g., a camera sensor), while another bead can correspond to the same object, but can be generated from data from another sensor type (e.g., a RADAR sensor, a LIDAR sensor, etc.).

[0042] The representation of the environment generated by the environment modeler 116 can be provided to the longitudinal velocity profile identifier 118 for selection of a longitudinal velocity profile. The longitudinal velocity profile can include acceleration and / or deceleration data that can be used by the lane trajectory determiner 124 and vehicle controls 126 of the autonomous vehicle 500 to help complete a lane change from the autonomous lane within the longitudinal space available to the ego vehicle 500 (e.g., a lane change gap) to the target lane. The longitudinal velocity profile identifier 118 can use the representation of the environment provided by the environment modeler 116 to identify one or more candidate lane change gaps in the target lane. Once one or more candidate gaps have been identified, the longitudinal velocity profile identifier can evaluate each candidate gap to determine a best candidate gap.

[0043] The longitudinal velocity profile identifier 118 can evaluate each of the candidate gaps in part by projecting a plurality (e.g., 100, 250, 700, 5000, etc.) of longitudinal velocity profile candidates to the target lane, which can be represented as a one-dimensional lane graph. In some embodiments, the longitudinal velocity profile identifier 118 can perform an evaluation of each of the plurality of longitudinal velocity profile candidates for each candidate gap. For a non-limiting example, if three candidate gaps are identified, then each of the plurality of longitudinal velocity profile candidates can be evaluated three times, once for each gap candidate.

[0044] The longitudinal velocity profile identifier 118 can evaluate (e.g., score, rank, rate, etc.) each longitudinal velocity profile candidate for each gap based on a set of criteria. The set of criteria can include classifications such as, but not limited to, speed adaptation duration, acceleration limit, deceleration limit, velocity vector field, fast movement limit, heuristic stop distance overlap, courtesy distance between vehicles, fallback gap count, speed loss, transition time, and / or safety checklist.

[0045] As used herein, the speed adaptation duration can be the amount of time required to adjust the speed of the ego vehicle 500. For example, the speed adaptation duration can be the amount of time for the ego vehicle 500 to decelerate and / or accelerate to align the ego vehicle 500 with a target gap candidate. The acceleration limit can refer to an upper limit on acceleration that the ego vehicle 500 cannot exceed. The deceleration limit can refer to a lower limit on acceleration that the ego vehicle 500 cannot fall below. The velocity vector field can refer to hard constraints formed by surrounding objects and lane geometry in the environment. For example, a lane change operation can be required to be performed before two lanes merge into one lane. A location in the environment where one lane ends (e.g., merges into another lane) can be defined as a boundary limit, and a lane change can be required to occur before the boundary limit is reached. The boundary limit can also include objects that constrain ego vehicle movement, such as additional vehicles in the environment. For example, a leading vehicle in the ego lane can define an upper limit on the ego’s speed such that any selected longitudinal velocity profile can not allow the ego to enter within a threshold distance of the leading vehicle. These boundary limits can be included in a vector field table that various components of the system 100 are to reference.

[0046] The fast movement limit can refer to a maximum rate of change of acceleration / deceleration allowed when performing a driving operation or maneuver by the ego vehicle. The heuristic stop distance can refer to an estimated distance required for the ego vehicle 500 (or another vehicle in the environment) to come to a complete stop at a given speed. Any overlap of stop distances (or 1D set forth) should be penalized as this would involve a potential collision in the future. The safety checklist can build such stop distance overlap information offline for quick reference at runtime. The courtesy distance between vehicles can refer to a threshold distance between vehicles established by a manufacturer, programmer, industry standard, and / or government agency, or otherwise to maintain a comfortable and / or safe driving experience for vehicle occupants. The count of fallback gaps can refer to the number of potential lane change gaps that the ego vehicle 500 can target if the initial target gap (e.g., the current gap immediately adjacent to the ego vehicle) is no longer available after the ego vehicle 500 initiates speed adaptation for the initial target gap. The speed loss can refer to the amount of speed that must be lost in order for the ego vehicle to align with a target gap candidate. The transition time can refer to the amount of time for the ego 500 to transition from an acceleration state to a deceleration state or from a deceleration state to an acceleration state.

[0047] In some embodiments, the longitudinal velocity profile identifier 118 can generate a standard score for each classification of each longitudinal velocity profile candidate for each gap. A cumulative score for each gap can then be generated based on a sum of the standard scores for each classification of each longitudinal velocity profile candidate. For example, where there are (but not limited to) 500 longitudinal velocity profile candidates, the same 500 candidates can be evaluated for each gap. The cumulative score for each gap can then be used to select a target gap for performing a lane change maneuver. For example, the gap with the best (e.g., highest, lowest, etc. depending on the scoring system) cumulative score can be selected as the target gap.

[0048] Once a target gap is selected, the longitudinal velocity profile identifier 118 can evaluate a plurality of longitudinal velocity profile candidates to determine a particular longitudinal velocity profile from the set or subset of longitudinal velocity profile candidates. For example, the plurality of longitudinal velocity profile candidates used to determine the target gap can then be reevaluated, in embodiments, the target gap determines a best longitudinal velocity profile from the plurality of longitudinal velocity profile candidates for performing a lane change maneuver into the target lane of the target gap. In some embodiments, each longitudinal lane change candidate can be evaluated again for the target gap, or in other embodiments, a subset of longitudinal lane change candidates can be considered for evaluation given the target gap. For example, where a subset is used, the subset can have been filtered from a larger group during the gap evaluation process. Thus, longitudinal velocity profile candidates that did not violate the velocity vector field (e.g., did not result in a collision of the ego vehicle with the ego lead object) and / or meet other criteria during the gap evaluation process that resulted in the ego vehicle 500 being in the target gap (e.g., not outside of the target gap, colliding with an object, etc.) can be used as the subset of longitudinal velocity profile candidates. In either example, the longitudinal velocity profile candidates evaluated for the target gap can be evaluated according to another set (e.g., different, same) of criteria (which can include some overlap criteria with the gap evaluation criteria) which can include a speed adaptation duration, an acceleration limit, a deceleration limit, a fast movement limit, a stop distance overlap, a courtesy distance, a back-off gap count, or an amount of speed loss to determine a velocity profile with a best score. Each longitudinal velocity profile candidate - or the subset thereof - can be scored based on the evaluation considering some or all of these criteria, and the longitudinal velocity profile candidate with the best (e.g., highest, lowest, etc. depending on the embodiment) score can be selected as the longitudinal velocity profile.

[0049] The longitudinal velocity profile identifier 118 can be bidirectionally communicatively coupled to the lateral path identifier 120 to exchange intermediate information to identify a lateral path profile that matches the selected longitudinal velocity profile. Similar to the process of identifying longitudinal velocity profile candidates, the lateral path profile identifier 120 can consider one or more criteria to evaluate a plurality of lateral path profile candidates— e.g., speed adaptation duration, acceleration limit, deceleration limit, fast movement limit, etc.— to identify and select a lateral path profile that has the best score.

[0050] The selected longitudinal velocity profile and the selected lateral path profile information can be passed to the lane change trajectory determiner 124 to determine commands for performing a lane change maneuver according to the identified profiles. For example, the lane change trajectory determiner 124 can determine and / or model a trajectory of the ego vehicle 500 within the environment based on the selected longitudinal velocity profile and the selected lateral path profile information. Using the trajectory, the lane change trajectory determiner 124 can determine / generate commands (e.g., control signals for steering, acceleration, braking, adjusting hover, etc. for the vehicle 500) that can be executed by the vehicle controls 126 to perform the lane change maneuver. The vehicle controls 126 can include, but are not limited to, operations for steering, acceleration, braking, adjusting hover, etc.

[0051] In some embodiments, once the longitudinal velocity profile and the target gap are determined, the longitudinal velocity profile identifier 118 can pass the longitudinal velocity profile information to the lane change trajectory determiner 124 to determine at least a portion of a lane change trajectory and corresponding commands according to the selected longitudinal velocity profile— e.g., by first decelerating and then accelerating to move the ego vehicle 500 toward the target lane. For example, the longitudinal velocity profile (or at least a portion thereof) can be first performed to align the ego vehicle with the target gap while remaining in the ego lane. In route to the target gap, the lateral path profile identifier 120 can evaluate a plurality of lateral profile candidates in view of the selected longitudinal velocity profile. For example, at each frame or iteration, the longitudinal velocity profile identifier 118 and the lateral path profile identifier 120 can exchange intermediate information with each other to identify the best lateral path profile. Information corresponding to the identified lateral path profile can then be passed to the lane change trajectory determiner 124. Once aligned, the identified lateral profile can be executed using the vehicle controls 126 based on information (e.g., commands) from the lane change trajectory determiner 124. Further, the identified lateral profile can be executed in conjunction with another longitudinal velocity profile or other portions of the initial longitudinal velocity profile to complete the lane change maneuver into the target lane.

[0052] Reference is now made to Figure 2 , Figure 2An example flowchart 200 for selecting a longitudinal velocity profile is depicted in accordance with some embodiments of the present disclosure. Figure 2 The environment block 202, the simple prediction block 204, the velocity profile search block 206, the evaluation block 208, and the lane change block 210 are included.

[0053] The environment block 202 can include environmental data - e.g. Figure 1 The model or representation of lanes and objects in the environment around the ego vehicle 500 described in the Background section. For example, the environmental data can include a model or representation of a road lane (e.g., based on high-definition map information) and the location of objects (e.g., vehicles, pedestrians, bicyclists, motorcyclists, etc.) within the lane.

[0054] Data from the environment block 202 can be passed to the simple prediction block 204 to generate a one-dimensional lane graph 212 representing objects in the ego lane and the target lane. The ID lane graph 212 can include a projection of the ego vehicle 500 (e.g., a forward simulation under constant speed and / or acceleration / deceleration assumptions) and longitudinal motion of environmental objects from both the ego lane and the target lane. For example, by determining the velocity, speed, acceleration, pose, heading, and / or other information corresponding to each object, a future position of each object at each time t over a period of time can be determined to effectively create a predictive model of the future lane change environment. As such, the prediction block 204 can process the environmental data to predict that an object corresponding to projection 214 can maintain a constant speed over a period of time. Further, the prediction block 204 can predict that another object corresponding to projection 216 will decelerate over a period of time.

[0055] The projections generated at the simple prediction block 204 can then be compared to the velocity vector field table to determine whether the predictions corresponding to each object in the environment and / or the ego vehicle 500 violate the velocity vector field. If a projection of an object is found to violate the velocity vector field, the velocity corresponding to the object can be clamped. In some embodiments, it can be assumed that a rear vehicle will not collide with a front (or ego) vehicle. Accordingly, the prediction of a rear vehicle that is predicted to collide with a front vehicle can be adjusted to account for the assumption that the following vehicle will likely decelerate so as not to collide with the front vehicle.

[0056] The 1D lane map 212 can then be passed from the simple prediction block 204 to the velocity distribution search block 206. In some embodiments, multiple (e.g., 100, 250, 700, 5000, etc.) longitudinal velocity distribution candidates can be projected onto the 1D lane map. Each of the multiple longitudinal velocity distribution candidates can then be evaluated to determine gap candidates. Taking into account gaps between objects, transition times, acceleration limits, rapid movement limits, and / or velocity vector field limits, a subset of the multiple longitudinal velocity distribution candidates can be excluded from consideration for selection as the optimal longitudinal velocity distribution, leaving only the longitudinal velocity distribution that achieves the lane change gap (e.g., candidate gap) between objects. For example, each of the longitudinal velocity distribution candidates 218A, 218B, and 218C can be retained for consideration because each of the longitudinal velocity distribution candidates 218A, 218B, and 218C lies between projections 214 and 216.

[0057] Information from the velocity distribution search block 206 can be passed to the evaluation block 208 to determine the optimal longitudinal velocity distribution among the longitudinal velocity distribution candidates. For example, each of the longitudinal velocity distribution candidates 218A, 218B, and 218C can be evaluated using a set of criteria to determine the optimal longitudinal velocity distribution. As discussed herein, this set of evaluation criteria may include, but is not limited to, classifications such as, but not limited to, speed adaptation duration, acceleration limits, deceleration limits, velocity vector field, rapid movement limits, heuristic stopping distance overlap, courtesy distance between vehicles, backlash count, speed loss, transition time, and / or safety checklists. For example, based on the evaluation of the criteria for each longitudinal velocity distribution candidate 218A, 218B, and 218C, longitudinal velocity distribution candidate 218B can be determined to be the optimal longitudinal velocity distribution 216 for the gap between projection 214 and projection. To interpret this result, it can be determined that longitudinal velocity distribution candidate 218A does not provide a courtesy distance between the ego vehicle 500 and the object corresponding to projection 214. Furthermore, it can be determined that the longitudinal velocity distribution candidate 218C will violate the components of the velocity vector field, because following the longitudinal velocity distribution candidate 218C may cause the ego vehicle 500 to violate the speed limit formed by the object corresponding to projection 216. Therefore, the longitudinal velocity distribution candidate 218B can be selected as the optimal longitudinal velocity distribution, and the information 218B corresponding to the longitudinal velocity distribution candidate can be passed to the lane change block 210.

[0058] Now for reference Figure 3A , Figure 3A A top view of an example lane-changing scenario 300A according to an embodiment of the present disclosure is depicted. In the exemplary lane-changing scenario 300A, based on receiving a command to perform a lane-changing maneuver, the system—for example, the self-vehicle 310A— Figure 1System 100 can generate an environmental model of its own vehicle 310A to perform lane change maneuvers. This model may include its own lane 350, a target lane 360, and a lane change period 370A. Furthermore, the model may include location data and predictions of the future positions of its own vehicle 310A, its leading vehicle 320A, its target following vehicle 330A, and its target leading vehicle 340A. The leading vehicle 320A, the target following vehicle 330A, the target leading vehicle 340A, and the lane change period 370A can establish constraints for lane change maneuvers, which may be included in a speed vector field table and / or a safety checklist.

[0059] Multiple longitudinal speed distribution candidates 312 (e.g., 100, 250, 700, 5000, etc.) can then be projected onto the target lane 360. Each of the multiple longitudinal speed distribution candidates can then be evaluated based on a set of criteria discussed herein to determine the target gap for the ego vehicle 310A. Although shown as a 2D representation in Figure 3, this is not intended to be restrictive. As described herein, the evaluation can be performed using a one-dimensional projection of the speed distribution. For example, candidate gaps may exist in the target lane 360 ​​(1) in front of the target leading vehicle 340A; (2) between the target trailing vehicle 330A and the target leading vehicle 340A; (3) behind the target trailing vehicle 330A. Each of the multiple longitudinal speed distribution candidates 312 can then be evaluated for each of these candidate gaps, and the target gap can then be selected using the cumulative score of each gap to perform a lane change maneuver into the target lane 360. Once the target gap is selected, the ego vehicle 310A can move to the appropriate position to perform the lane change maneuver, as discussed herein and further regarding Figure 3B .

[0060] Now for reference Figure 3B , Figure 3B A one-dimensional lane change scenario 300B according to an embodiment of the present disclosure is depicted. The 1D lane change scenario 300B is similar in some respects to... Figure 3A The example lane-changing scenario 300A includes a 1D lane map 380, an ego vehicle 310B, an ego-leading vehicle 320B, a target following vehicle 330B, and a target leading vehicle 340B. As discussed herein, the system can project the longitudinal motion of the ego vehicle 310A and the ego-leading vehicle 320A from the ego lane 350 to the target lane 360. Figure 3A The self-lane 350 and the target lane 360 ​​are compressed to such a size Figure 3BThe 1D lane change scenario 300B is shown with a 1D lane graph 380. The 1D lane change scenario 300B also includes a trajectory projection 332 corresponding to a predicted trajectory for the target following car 330B, a trajectory projection 334 corresponding to a predicted trajectory of the target leading car 340B, a trajectory projection 336 corresponding to a predicted trajectory of the ego leading car 320, and a longitudinal path 338 representing a selected longitudinal path to move the ego vehicle 310B to a position 372 to perform a lane change operation.

[0061] In operation, the ego vehicle 310B can remain in the ego lane and perform a deceleration command from the longitudinal path 338 to position the ego vehicle 310B between the target following car 330B and the target leading car 340B. At a position 370, which can correspond to a time increment (e.g., t = 4s) after the deceleration command begins, the ego vehicle 310B can transition to perform an acceleration command from the longitudinal path 338 to move the ego vehicle 310B to a position 372 to perform a lateral lane change maneuver. At the position 372, the ego vehicle 310B can perform a lateral lane change maneuver to move from the ego lane to the target lane.

[0062] Referring now to Figure 4 Each block of the method 400 described herein includes a computational process that can be performed using any combination of hardware, firmware, and / or software. For instance, various functions can be carried out by a processor executing instructions stored in memory. The method 400 can also be embodied as computer-usable instructions stored on computer storage media. The method 400 can be provided by a standalone application, a service or hosted service (standalone or in combination with another hosted service), or a plug-in to another product, to name a few. Figure 1 The method 400 is described with respect to the system 100. However, the method can additionally or alternatively be performed by any one system or any combination of systems, including but not limited to those described herein.

[0063] Figure 4 is a flowchart illustrating a method 400 for identifying a longitudinal velocity distribution, in accordance with some embodiments of the present disclosure. At block B402, the method 400 includes generating a representation of an environment surrounding an ego vehicle. In some embodiments, the representation can indicate at least one ego lane, a target lane adjacent to the ego lane, and a location of one or more objects within at least one of the ego lane or the target lane. For example, outputs from the object detector 112 and / or the lane identifier 114 can be provided to the environment modeler 116, which can generate a model or representation of the lanes of the roadway and the locations of objects within the lanes.

[0064] At block B404, the method 400 includes identifying one or more candidate gaps within the target lane for the ego vehicle to maneuver into based at least in part on the representation of the environment. In some embodiments, each of the one or more candidate gaps is at least partially delineated by a location of an object of one or more objects within the representation of the environment. For example, upon receiving a lane change command, the longitudinal velocity profile identifier can use the representation of the environment to identify one or more candidate gaps in the target lane, which can then be evaluated.

[0065] At block B406, the method 400 includes identifying a target gap of the one or more candidate gaps based at least in part on evaluating a plurality of longitudinal velocity profile candidates for each of the one or more candidate gaps with respect to one or more first criteria. For example, each longitudinal velocity profile candidate can be evaluated (e.g., scored) for each gap based on a set of criteria (e.g., speed adaptation duration, acceleration limit, deceleration limit, velocity vector field, fast moving limit, etc.) to generate an accumulated score for each gap. The accumulated score for each gap can then be used to select the target gap for the current lane change maneuver.

[0066] At block B408, the method 400 includes identifying a longitudinal velocity profile from the plurality of longitudinal velocity profile candidates for the target gap based at least in part on evaluating the plurality of longitudinal velocity profile candidates with respect to one or more second criteria. For example, the longitudinal velocity profile candidates evaluated for the target gap can be evaluated according to another set (e.g., different in embodiments) of criteria (which can include some overlapping criteria with the gap evaluation criteria), which can include speed adaptation duration, acceleration limit, deceleration limit, fast moving limit, stop distance overlap, courtesy distance, back-off gap count, or amount of speed loss, to determine a velocity profile with the best score. Each longitudinal velocity profile candidate - or a subset thereof - can be scored based on an evaluation that takes into account some or all of these criteria, and the longitudinal velocity profile candidate with the best (e.g., highest, lowest, etc., depending on embodiments) score can be selected as the longitudinal velocity profile.

[0067] At block B410, the method 400 performs the lane change according to the longitudinal velocity profile. For example, once the longitudinal velocity profile and the target gap are determined, the system can perform the lane change maneuver according to the longitudinal velocity profile - e.g., first accelerating and then decelerating, etc. In addition to the longitudinal velocity profile, a lateral path or trajectory profile can also be evaluated and selected to define the lateral motion of the lane change maneuver.

[0068] Example autonomous vehicle

[0069] Figure 5AA diagram of an example autonomous vehicle 500 in accordance with some embodiments of the present disclosure. Autonomous vehicle 500 (alternatively referred to herein as “vehicle 500”) can include, but is not limited to, a passenger vehicle such as a car, truck, bus, ambulance, shuttle, electric or motorized bicycle, motorcycle, fire truck, police car, ambulance, boat, construction vehicle, underwater vessel, drone, and / or other type of vehicle (e.g., unmanned and / or capable of accommodating one or more passengers). Autonomous vehicles are often described in terms of levels of automation as defined by a division of the United States Department of Transportation, the National Highway Traffic Safety Administration (NHTSA), and the Society of Automotive Engineers (SAE) “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles” (Standard No. J3016-201806 published June 15, 2018, Standard No. J3016-201609 published September 30, 2016, and prior and future versions of this standard). Vehicle 500 can be capable of implementing functionality that complies with one or more of Levels 3-5 of autonomous driving. For example, depending on the embodiment, vehicle 500 can be capable of implementing conditional automation (Level 3), high automation (Level 4), and / or full automation (Level 5).

[0070] Vehicle 500 can include components such as a chassis, a body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other components of a vehicle. Vehicle 500 can include a propulsion system 550 such as an internal combustion engine, a hybrid electric power plant, a fully electric motor, and / or another type of propulsion system. Propulsion system 550 can be connected to a drivetrain of vehicle 500 that can include a transmission in order to effect propulsion of vehicle 500. Propulsion system 550 can be controlled in response to receiving a signal from a throttle / accelerator 552.

[0071] A steering system 554, which can include a steering wheel, can be used to steer vehicle 500 (e.g., along a desired path or route) while propulsion system 550 is operating (e.g., while the vehicle is in motion). Steering system 554 can receive a signal from a steering actuator 556. For full automation (Level 5) functionality, a steering wheel can be optional.

[0072] A braking sensor system 546 can be used to operate vehicle brakes in response to receiving a signal from a brake actuator 548 and / or a brake sensor.

[0073] A computing system 502 can include one or more system on chips (SoCs) 504 Figure 5C) and / or one or more controllers 536 of one or more GPUs can provide signals (e.g., representative of commands) to one or more components and / or systems of the vehicle 500. For example, the one or more controllers can send signals to operate vehicle brakes via one or more brake actuators 548, to operate a steering system 554 via one or more steering actuators 556, to operate a propulsion system 550 via one or more throttle / accelerator 552. The one or more controllers 536 can include one or more on-board (e.g., integrated) computing devices (e.g., supercomputers) that process sensor signals and output operational commands (e.g., signals representative of commands) to enable autonomous driving and / or to assist a human driver in driving the vehicle 500. The one or more controllers 536 can include a first controller 536 for autonomous driving functions, a second controller 536 for functional safety functions, a third controller 536 for artificial intelligence functions (e.g., computer vision), a fourth controller 536 for infotainment functions, a fifth controller 536 for redundancy in emergency situations, and / or other controllers. In some examples, a single controller 536 can handle two or more of the above functions, two or more controllers 536 can handle a single function, and / or any combination thereof.

[0074] The one or more controllers 536 can provide signals for controlling one or more components and / or systems of the vehicle 500 in response to sensor data (e.g., sensor inputs) received from one or more sensors. The sensor data can be received from, for example and without limitation, a global navigation satellite system sensor 558 (e.g., a global positioning system sensor), a RADAR sensor 560, an ultrasonic sensor 562, a LIDAR sensor 564, an inertial measurement unit (IMU) sensor 566 (e.g., an accelerometer, a gyroscope, a magnetic compass, a magnetometer, etc.), a microphone 596, a stereo camera 568, a wide-angle camera 570 (e.g., a fisheye camera), an infrared camera 572, a surround camera 574 (e.g., a 360-degree camera), a long-range and / or mid-range camera 598, a speed sensor 544 (e.g., to measure a speed of the vehicle 500), a vibration sensor 542, a steering sensor 540, a brake sensor (e.g., as part of a brake sensor system 546), and / or other sensor types.

[0075] One or more of the controllers 536 can receive input (e.g., represented by input data) from an instrument cluster 532 of the vehicle 500 and provide output (e.g., represented by output data, display data, etc.) via a human-machine interface (HMI) display 534, an audible annunciator, a speaker, and / or via other components of the vehicle 500. These outputs can include, for example and without limitation, vehicle speed, speed, time, map data (e.g., represented by a map display), and / or other information. The one or more controllers 536 can also receive input from and provide output to other components of the vehicle 500, such as the one or more sensors, the one or more actuators, and / or other components. Figure 5Cinformation such as information about objects and object states as perceived by the controller 536, and so on. For example, the HMI display 534 can display information about the presence of one or more objects (e.g., a street sign, a warning sign, a traffic light change, and so on) and / or information about driving maneuvers that the vehicle has made, is making, or will make (e.g., now lane change, exit 34B in two miles, and so on).

[0076] The vehicle 500 further includes a network interface 524 that can communicate over one or more networks using one or more wireless antennas 526 and / or modems. For example, the network interface 524 can be capable of communicating over LTE, WCDMA, UMTS, GSM, CDMA2000, and so on. The one or more wireless antennas 526 can also enable communication between objects (e.g., vehicles, mobile devices, and so on) in the implementation environment using one or more local area networks such as Bluetooth, Bluetooth LE, Z-Wave, ZigBee, and so on and / or one or more low power wide area networks (LPWANs) such as LoRaWAN, SigFox, and so on.

[0077] Figure 5B Examples of camera positions and fields of view of the example autonomous vehicle 500 according to some embodiments of the present disclosure. Figure 5A Examples of camera positions and fields of view of the example autonomous vehicle 500 according to some embodiments of the present disclosure.

[0078] Camera types for the cameras can include, but are not limited to, digital cameras that can be suitable for use with components and / or systems of the vehicle 500. The cameras can operate at Automotive Safety Integrity Level (ASIL) B and / or at another ASIL. The camera types can have any image capture rate, such as 60 frames per second (fps), 120 fps, 240 fps, and so on, depending on the embodiment. The cameras can be capable of using a rolling shutter, a global shutter, another type of shutter, or a combination thereof. In some examples, a color filter array can include a red- white-white-white (RCCC) color filter array, a red-white-white-blue (RCCB) color filter array, a red-blue-green-white (RBGC) color filter array, a Foveon X3 color filter array, a Bayer sensor (RGGB) color filter array, a monochrome sensor color filter array, and / or another type of color filter array. In some embodiments, clear pixel cameras such as cameras with a

[0079] In some examples, one or more of the cameras can be used to perform advanced driver assistance system (ADAS) functions (e.g., as part of a redundant or fail-safe design). For example, a multi-function monocular camera can be installed to provide functions including lane departure warning, traffic sign assist, and intelligent headlamp control. One or more of the cameras (e.g., all of the cameras) can simultaneously record and provide image data (e.g., video).

[0080] One or more of the cameras can be mounted in mounting assemblies such as custom designed (3-D printed) assemblies to cut off stray light and reflections from within the car that can interfere with the image data capture capabilities of the cameras (e.g., reflections from the dashboard reflected in the windshield mirror). With respect to wing mirror mounting assemblies, the wing mirror assemblies can be custom 3-D printed such that the camera mounting plate matches the shape of the wing mirror. In some examples, one or more cameras can be integrated into the wing mirror. For side view cameras, one or more cameras can also be integrated into the four pillars at each corner of the cab.

[0081] Cameras with fields of view that include the portion of the environment in front of the vehicle 500 (e.g., front-facing cameras) can be used for surround view to help identify the forward path and obstacles, and to assist in providing information critical to generating an occupancy grid and / or determining a preferred vehicle path with the help of one or more controllers 536 and / or control SoCs. Front-facing cameras can be used to perform many of the same ADAS functions as LIDAR, including emergency braking, pedestrian detection, and collision avoidance. Front-facing cameras can also be used for ADAS functions and systems including lane departure warning ("LDW"), adaptive cruise control ("ACC"), and / or other functions such as traffic sign recognition.

[0082] A wide variety of cameras can be used in the front-facing configuration, including, for example, monocular camera platforms including CMOS (complementary metal-oxide semiconductor) color imagers. Another example can be a wide-angle camera 570, which can be used to perceive objects (e.g., pedestrians, intersection traffic, or bicycles) entering the field of view from the periphery. Although Figure 5B Although only one wide-angle camera is illustrated in FIG. 5, there can be any number of wide-angle cameras 570 on the vehicle 500. In addition, long-range cameras 598 (e.g., long-view stereo camera pairs) can be used for depth-based object detection, especially for objects for which a neural network has not been trained. Long-range cameras 598 can also be used for object detection and classification and basic object tracking.

[0083] One or more stereo cameras 568 can also be included in the front-facing configuration. Stereo cameras 568 can include an integrated control unit that includes a scalable processing unit that can provide a multi-core microprocessor with integrated CAN or Ethernet interface and programmable logic (FPGA) on a single chip. Such a unit can be used to generate a 3-D map of the vehicle's environment, including distance estimates for all points in the image. Alternative stereo cameras 568 can include a compact stereo vision sensor that can include two camera lenses (one on the left and one on the right) and an image processing chip that can measure the distance from the vehicle to a target object and use the generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning functions. Other types of stereo cameras 568 can be used in addition to or instead of those described herein.

[0084] Cameras with fields of view that include portions of the environment to the side of vehicle 500 (e.g., side-view cameras) can be used for surround view, providing information used to create and update the occupancy grid and to generate side-crash collision warnings. For example, surround cameras 574 (e.g., four surround cameras 574 as shown in FIG. 6B) can be placed on vehicle 500. Surround cameras 574 can include wide-view cameras 570, fisheye cameras, 360-degree cameras, and / or the like. In one example, four fisheye cameras can be placed on the front, back, and sides of the vehicle. In an alternative arrangement, a vehicle can use three surround cameras 574 (e.g., left, right, and back) and can utilize one or more other cameras (e.g., a forward-facing camera) as a fourth surround view camera. Figure 5B

[0085] Cameras with fields of view that include portions of the environment behind vehicle 500 (e.g., rear-view cameras) can be used for assist parking, surround view, rear collision warnings, and to create and update the occupancy grid. A wide variety of cameras can be used, including but not limited to cameras that are also suitable as front-facing cameras as described herein (e.g., long- and / or mid-range cameras 598, stereo cameras 568, infrared cameras 572, etc.).

[0086] Figure 5C For use in a vehicle according to some embodiments of the present disclosure Figure 5A ​FIG. 1 is a block diagram of an example system architecture of an example autonomous vehicle 500. It should be understood that this arrangement and other arrangements described herein are set forth merely as examples. Other arrangements and elements (e.g., machines, interfaces, functions, orders, groupings of functions, etc.) can be used in addition to or instead of those shown, and some elements can be wholly omitted. Further, many of the elements described herein are functional entities that can be implemented as discrete or distributed components or in conjunction with other components, and in any suitable combinations and locations. Various functions described herein as being performed by an entity can be implemented in hardware, firmware, and / or software. For instance, various functions can be implemented by a processor executing instructions stored in a memory.

[0087] Figure 5C Each of the components, features, and systems of vehicle 500 are illustrated as being connected via a bus 502. Bus 502 can include a controller area network (CAN) data interface (alternatively referred to herein as a "CAN bus"). The CAN can be a network within vehicle 500 that is used to assist in controlling various features and functions of vehicle 500, such as the actuation of brakes, acceleration, braking, steering, windshield wipers, etc. The CAN bus can be configured to have tens or even hundreds of nodes, each with its own unique identifier (e.g., CAN ID). The CAN bus can be read to find steering wheel angle, ground speed, revolutions per minute (RPM) of the engine, button positions, and / or other vehicle status indicators. The CAN bus can be ASIL B compliant.

[0088] Although bus 502 is described herein as a CAN bus, this is not intended to be limiting. For example, FlexRay and / or Ethernet can be used in addition to or instead of a CAN bus. Further, although bus 502 is represented with a single line, this is not intended to be limiting. For example, there can be any number of buses 502, which can include one or more CAN buses, one or more FlexRay buses, one or more Ethernet buses, and / or one or more other types of buses that use different protocols. In some examples, two or more buses 502 can be used to perform different functions, and / or can be used for redundancy. For example, a first bus 502 can be used for collision avoidance functions, and a second bus 502 can be used for drive control. In any example, each bus 502 can communicate with any component of vehicle 500, and two or more buses 502 can communicate with the same components. In some examples, each SoC 504, each controller 536, and / or each computer within the vehicle can have access to the same input data (e.g., inputs from sensors of vehicle 500), and can be connected to a common bus, such as a CAN bus.

[0089] Vehicle 500 may include one or more controllers 536, such as those described herein. Figure 5A The controllers described herein. Controller 536 can be used for a wide variety of functions. Controller 536 can be coupled to any other different components and systems of vehicle 500 and can be used for the control of vehicle 500, artificial intelligence of vehicle 500, infotainment and / or the like for vehicle 500.

[0090] Vehicle 500 may include one or more System-on-Chip (SoC) 504s. SoC 504 may include a CPU 506, GPU 508, processor 510, cache 512, accelerator 514, data storage 516, and / or other components and features not shown. SoC 504 can be used to control vehicle 500 across a wide variety of platforms and systems. For example, one or more SoCs 504s may be combined with an HD map 522 in a system (e.g., the system of vehicle 500), the HD map being transmitted via a network interface 524 from one or more servers (e.g., [server name missing]). Figure 5D One or more servers (578) receive map refresh and / or updates.

[0091] CPU 506 may include CPU clusters or CPU complexes (alternatively referred to herein as "CCPLEX"). CPU 506 may include multiple cores and / or L2 cache. For example, in some embodiments, CPU 506 may include eight cores in a coherent multiprocessor configuration. In some embodiments, CPU 506 may include four dual-core clusters, each with a dedicated L2 cache (e.g., 2MB L2 cache). CPU 506 (e.g., CCPLEX) may be configured to support simultaneous cluster operation, such that any combination of clusters of CPU 506 can be active at any given time.

[0092] CPU 506 can implement power management capabilities including one or more of the following features: automatic clock gating of hardware blocks when idle to conserve dynamic power; clock gating of each core when the core is not actively executing instructions due to the execution of WFI / WFE instructions; independent power gating of each core; independent clock gating of each core cluster when all cores are clock-gated or power-gated; and / or independent power gating of each core cluster when all cores are power-gated. CPU 506 can further implement enhanced algorithms for managing power states, wherein allowed power states and desired wake-up times are specified, and the hardware / microcode determines the optimal power state to enter for the core, cluster, and CCPLEX. The processing core can support simplified power state entry sequences in software, with this work offloaded to the microcode.

[0093] GPU 508 can include an integrated GPU (alternatively referred to herein as an “iGPU”). GPU 508 can be programmable and efficient for parallel workloads. In some examples, GPU 508 can use an enhanced tensor instruction set. GPU 508 can include one or more streaming microprocessors, where each streaming microprocessor can include an LI cache (e.g., an LI cache having at least 96 KB of storage capacity), and two or more of the streaming microprocessors can share an L2 cache (e.g., an L2 cache having 512 KB of storage capacity). In some embodiments, GPU 508 can include at least eight streaming microprocessors. GPU 508 can use a compute application programming interface (API). Additionally, GPU 508 can use one or more parallel computing platforms and / or programming models (e.g., NVIDIA’s CUDA).

[0094] In the case of automotive and embedded uses, GPU 508 can be power-optimized for best performance. For example, GPU 508 can be fabricated on a fin field-effect transistor (FinFET). However, this is not intended to be limiting, and GPU 508 can be fabricated using other semiconductor fabrication processes. Each streaming microprocessor can incorporate several mixed-precision processing cores divided into multiple blocks. For example, and without limitation, 64 PF32 cores and 32 PF64 cores can be divided into four processing blocks. In such an example, each processing block can be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed-precision NVIDIA Tensor cores for deep learning matrix arithmetic, an L0 instruction cache, a thread warp scheduler, a dispatch unit, and / or a 64 KB register file. Additionally, the streaming microprocessor can include independent parallel integer and floating point data paths to provide efficient execution of workloads with a mix of compute and address compute. The streaming microprocessor can include independent thread scheduling capabilities to allow for more fine-grained synchronization and cooperation between parallel threads. The streaming microprocessor can include a combined LI data cache and shared memory unit to improve performance while simplifying programming.

[0095] GPU 508 can include a high-bandwidth memory (HBM) and / or a 16 GB HBM2 memory subsystem that provides approximately 900 GB / s of peak memory bandwidth in some examples. In some examples, in addition to or alternatively from HBM memory, a synchronous graphics random access memory (SGRAM) can be used, such as a fifth generation graphics double data rate synchronous random access memory (GDDR5).

[0096] GPU 508 can include a unified memory technology that includes access counters to allow memory pages to be migrated more precisely to the processors that access them most frequently, thereby improving efficiency of memory ranges shared between processors. In some examples, address translation services (ATS) support can be used to allow GPU 508 to directly access CPU 506 page tables. In such examples, when a GPU 508 memory management unit (MMU) experiences a miss, an address translation request can be transmitted to CPU 506. In response, CPU 506 can look up a virtual-to-physical mapping for the address in its page tables and transmit the translation back to GPU 508. In this way, the unified memory technology can allow a single unified virtual address space for memory of both CPU 506 and GPU 508, thereby simplifying GPU 508 programming and porting applications to GPU 508.

[0097] Further, GPU 508 can include access counters that can track how frequently GPU 508 accesses other processors’ memory. The access counters can help ensure that memory pages are migrated to the physical memory of the processor that accesses these pages most frequently.

[0098] SoC 504 can include any number of caches 512, including those described herein. For example, caches 512 can include an L3 cache available to both CPU 506 and GPU 508 (e.g., connected to both CPU 506 and GPU 508). Caches 512 can include a write-back cache that can track the state of a line, for example, by using a cache coherency protocol (e.g., MEI, MESI, MSI, etc.). Depending on the embodiment, the L3 cache can include 4 MB or more, although smaller cache sizes can also be used.

[0099] SoC 504 can include one or more arithmetic logic units (ALUs) that can be used to perform processing with respect to any of a variety of tasks or operations of vehicle 500, such as processing a DNN. Further, SoC 504 can include a floating point unit (FPU) or other mathematical co-processor or digital co-processor type for performing mathematical operations within the system. For example, SoC 104 can include one or more FPUs integrated as execution units within CPU 506 and / or GPU 508.

[0100] The SoC 504 can include one or more accelerators 514 (e.g., hardware accelerators, software accelerators, or a combination thereof). For example, the SoC 504 can include a hardware acceleration cluster that can include optimized hardware accelerators and / or a large on-chip memory. This large on-chip memory (e.g., 4 MB SRAM) can enable the hardware acceleration cluster to accelerate neural networks and other computations. The hardware acceleration cluster can be used to supplement the GPU 508 and offload some of the tasks of the GPU 508 (e.g., freeing up more cycles of the GPU 508 for performing other tasks). As one example, the accelerators 514 can be used for targeted workloads (e.g., perception, convolutional neural networks (CNNs), etc.) that are stable enough to accelerate easily. As used herein, the term “CNN” can include all types of CNNs, including region-based or region convolutional neural networks (RCNNs) and fast RCNNs (e.g., for object detection).

[0101] The accelerators 514 (e.g., hardware acceleration cluster) can include a deep learning accelerator (DLA). The DLA can include one or more tensor processing units (TPUs) that can be configured to provide an additional 100 trillion operations per second for deep learning applications and inferencing. The TPU can be an accelerator that is configured to perform and optimized for performing image processing functions (e.g., for CNNs, RCNNs, etc.). The DLA can be further optimized for a specific set of neural network types and floating point operations and inferencing. The design of the DLA can provide higher performance per mm than general purpose GPUs and far exceeds the performance of CPUs. The TPU can perform several functions, including single instance convolution functions, support for INT8, INT16, and FP16 data types for both features and weights, for example, and post-processor functions.

[0102] The DLA can perform neural networks, especially CNNs, on processed or unprocessed data for any of a wide variety of functions, such as and not limited to: CNNs for object recognition and detection using data from a camera sensor; CNNs for distance estimation using data from a camera sensor; CNNs for emergency vehicle detection and identification and detection using data from a microphone; CNNs for face recognition and vehicle owner identification using data from a camera sensor; and / or CNNs for safety and / or safety related events.

[0103] The DLA can perform any of the functions of the GPU 508, and by using an inferencing accelerator, the designer can target the DLA or the GPU 508 for any function. For example, the designer can focus the processing and floating point operations of the CNNs on the DLA and leave other functions to the GPU 508 and / or other accelerators 514.

[0104] Accelerator 514 (e.g., hardware acceleration cluster) can include a programmable vision accelerator (PVA), which can be alternatively referred to herein as a computer vision accelerator. The PVA can be designed and configured to accelerate computer vision algorithms for advanced driver assistance systems (ADAS), autonomous driving, and / or augmented reality (AR) and / or virtual reality (VR) applications. The PVA can provide a balance between performance and flexibility. For example, each PVA can include, for example and without limitation, any number of reduced instruction set computer (RISC) cores, direct memory access (DMA), and / or any number of vector processors.

[0105] The RISC cores can interact with image sensors (e.g., image sensors of any of the cameras described herein), image signal processors, and / or the like. Each of the RISC cores can include any number of memories. Depending on the embodiment, the RISC cores can use any of several protocols. In some examples, the RISC cores can execute a real-time operating system (RTOS). The RISC cores can be implemented using one or more integrated circuit devices, application specific integrated circuits (ASICs), and / or memory devices. For example, the RISC cores can include an instruction cache and / or a tightly coupled RAM.

[0106] The DMA can enable components of the PVA to access system memory independently of the CPU 506. The DMA can support any number of features to provide optimization to the PVA, including, but not limited to, supporting multi-dimensional addressing and / or circular addressing. In some examples, the DMA can support addressing up to six or more dimensions, which can include block width, block height, block depth, horizontal block stride, vertical block stride, and / or depth stride.

[0107] The vector processors can be programmable processors that can be designed to efficiently and flexibly execute programming for computer vision algorithms and provide signal processing capabilities. In some examples, the PVA can include a PVA core and two vector processing subsystem partitions. The PVA core can include a processor subsystem, one or more DMA engines (e.g., two DMA engines), and / or other peripherals. The vector processing subsystems can operate as the main processing engines of the PVA and can include a vector processing unit (VPU), an instruction cache, and / or a vector memory (e.g., VMEM). The VPU core can include a digital signal processor, such as, for example, a single instruction multiple data (SIMD), very long instruction word (VLIW) digital signal processor. The combination of SIMD and VLIW can enhance throughput and rate.

[0108] Each of the vector processors can include an instruction cache and can be coupled to a dedicated memory. As a result, in some examples, each of the vector processors can be configured to execute independently of the other vector processors. In other examples, the vector processors included in a particular PVA can be configured to employ data parallelization. For example, in some embodiments, multiple vector processors included in a single PVA can execute the same computer vision algorithm, but on different regions of an image. In other examples, the vector processors included in a particular PVA can execute different computer vision algorithms on the same image simultaneously, or even different algorithms on sequential images or portions of an image. Any number of PVAs can be included in the hardware acceleration cluster, and any number of vector processors can be included in each of the PVAs, among other things. Furthermore, the PVAs can include additional error-correcting code (ECC) memory to enhance overall system security.

[0109] The accelerator 514 (e.g., hardware acceleration cluster) can include an on-chip computer vision network and SRAM to provide high bandwidth, low latency SRAM for the accelerator 514. In some examples, the on-chip memory can include at least 4 MB of SRAM composed of, for example and without limitation, eight field-programmable memory blocks, which can be accessed by both the PVA and the DLA. Each pair of memory blocks can include an advanced peripheral bus (APB) interface, configuration circuitry, a controller, and a multiplexer. Any type of memory can be used. The PVA and the DLA can access the memory via a backbone that provides high-speed memory access to the PVA and the DLA. The backbone can include an on-chip computer vision network that interconnects the PVA and the DLA to the memory, for example using an APB.

[0110] The on-chip computer vision network can include an interface that determines that both the PVA and the DLA provide ready and valid signals before transmitting any control signals / addresses / data. Such an interface can provide separate phases and separate channels for transmitting control signals / addresses / data, as well as burst communications for continuous data transmission. This type of interface can comply with ISO 26262 or IEC 61508 standards, but other standards and protocols can also be used.

[0111] In some examples, the SoC 504 can include a real-time ray tracing hardware accelerator, such as described in U.S. Patent Application No. 16 / 101,232, filed August 10, 2018. The real-time ray tracing hardware accelerator can be used to quickly and efficiently determine locations and extents of objects (e.g., within a world model) in order to generate real-time visualizations simulations for RADAR signal interpretation, for sound propagation synthesis and / or analysis, for SONAR system simulation, for general wave propagation simulation, for comparison with LIDAR data for purposes of localization and / or other functionality, and / or for other uses. In some embodiments, one or more tree traversal units (TTUs) can be used to perform one or more ray tracing related operations.

[0112] The accelerator 514 (e.g., a hardware accelerator cluster) has a wide range of autonomous driving uses. The PVA can be a programmable vision accelerator that can be used for key processing stages in ADAS and autonomous vehicles. The capabilities of the PVA are a good match for algorithm domains that require predictable processing, low power, and low latency. In other words, the PVA performs well on semi-dense or dense regular computations, and even on small data sets that require predictable runtimes with low latency and low power. Thus, in the context of a platform for autonomous vehicles, the PVA is designed to run classical computer vision algorithms because they are effective at object detection and integer math operations.

[0113] For example, according to one embodiment of the technology, the PVA is used to perform computer stereo vision. In some examples, a semi-global matching based algorithm can be used, although this is not intended to be limiting. Many applications for level 3-5 autonomous driving require instant motion estimation / stereo matching (e.g., structure from motion, pedestrian recognition, lane detection, etc.). The PVA can perform computer stereo vision functions on input from two monocular cameras.

[0114] In some examples, the PVA can be used to perform dense optical flow. Raw RADAR data is processed according to a process (e.g., using a 4D fast Fourier transform) to provide processed RADAR. In other examples, the PVA is used for time-of-flight depth processing, such as by processing raw time-of-flight data to provide processed time-of-flight data.

[0115] The DLA can be used to run any type of network to enhance control and driving safety, including, for example, a neural network that outputs a confidence metric for each object detection. Such a confidence value can be interpreted as a probability, or as providing a relative "weight" for each detection compared to other detections. The confidence value enables the system to make further decisions about which detections should be considered true positive detections and not false positive detections. For example, the system can set a threshold for confidence, and only consider detections that exceed the threshold as true positive detections. In an automatic emergency braking (AEB) system, false positive detections would cause the vehicle to automatically perform an emergency brake, which is obviously undesirable. Thus, only the most confident detections should be considered a trigger for AEB. The DLA can run a neural network for regression of a confidence value. The neural network can take as its input at least some subset of parameters, such as a bounding box dimension, a ground plane estimate obtained (e.g., from another subsystem), inertial measurement unit (IMU) sensor 566 outputs related to vehicle 500 orientation, distance, 3D position estimates of objects obtained from the neural network and / or other sensors (e.g., LIDAR sensor 564 or RADAR sensor 560), etc.

[0116] SoC 504 can include one or more data stores 516 (e.g., memory). Data stores 516 can be on-chip memory of SoC 504, which can store neural networks to be executed on the GPU and / or DLA. In some examples, for redundancy and safety, data stores 516 can be large enough in capacity to store multiple instances of a neural network. Data stores 512 can include L2 or L3 cache 512. References to data stores 516 can include references to memory associated with PVAs, DLAs, and / or other accelerators 514 as described herein.

[0117] The SoC 504 can include one or more processors 510 (e.g., embedded processors). The processors 510 can include a boot and power management processor, which can be a specialized processor and subsystem for handling boot power and management functions and related security implementations. The boot and power management processor can be part of the SoC 504 boot sequence and can provide run-time power management services. The boot power and management processor can provide clock and voltage programming, auxiliary system low power state transitions, SoC 504 thermal and temperature sensor management, and / or SoC 504 power state management. Each temperature sensor can be implemented as a ring oscillator whose output frequency is proportional to temperature, and the SoC 504 can use the ring oscillator to detect the temperature of the CPU 506, GPU 508, and / or accelerator 514. If it is determined that the temperature exceeds a threshold, the boot and power management processor can enter a temperature fault routine and place the SoC 504 in a lower power state and / or place the vehicle 500 in a driver safe park mode (e.g., safely park the vehicle 500).

[0118] The processors 510 can further include a set of embedded processors that can be used as an audio processing engine. The audio processing engine can be an audio subsystem that allows for full hardware support for multi-channel audio over multiple interfaces and a range of widely flexible audio I / O interfaces. In some examples, the audio processing engine is a specialized processor core with a digital signal processor with dedicated RAM.

[0119] The processors 510 can further include an always-on processor engine that can provide the necessary hardware features to support low power sensor management and wake-up use cases. The always-on processor engine can include a processor core, tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I / O controller peripherals, and routing logic.

[0120] The processors 510 can further include a security cluster engine that includes a specialized processor subsystem that handles security management for automotive applications. The security cluster engine can include two or more processor cores, tightly coupled RAM, supporting peripherals (e.g., timers, interrupt controllers, etc.), and / or routing logic. In a secure mode, the two or more cores can operate in a lockstep mode and act as a single core with comparison logic that detects any differences between their operations.

[0121] The processors 510 can further include a real-time camera engine that can include a specialized processor subsystem for handling real-time camera management.

[0122] The processor 510 can further include a high dynamic range signal processor, which can include an image signal processor, which is a hardware engine that is part of the camera processing pipeline.

[0123] The processor 510 can include a video image compositor, which can be a processing block (e.g., implemented on a microprocessor), that implements video post-processing functions needed by the video playback application to produce the final image for the player window. The video image compositor can perform lens distortion correction on the wide-angle camera 570, surround camera 574, and / or on the cab-in monitor camera sensors. The cab-in monitor camera sensors are preferably monitored by a neural network running on another instance of the advanced SoC, configured to recognize cab-in events and respond accordingly. The cab-in system can perform lip reading to activate mobile phone services and place a call, dictate an email, change the vehicle destination, activate or change the vehicle's infotainment system and settings, or provide voice-activated web surfing. Certain functions are only available to the driver when the vehicle is operating in autonomous mode, and are disabled otherwise.

[0124] The video image compositor can include enhanced temporal noise reduction for spatial and temporal noise reduction. For example, where motion is present in the video, the noise reduction appropriately weights the spatial information, reducing the weight of information provided by neighboring frames. Where the image or portions of the image do not include motion, the temporal noise reduction performed by the video image compositor can use information from previous images to reduce noise in the current image.

[0125] The video image compositor can also be configured to perform stereo correction on input stereo lens frames. The video image compositor can further be used for user interface composition when the operating system desktop is in use and the GPU 508 does not need to continuously render new surfaces. Even when the GPU 508 is powered on and active, doing 3D rendering, the video image compositor can be used to offload the GPU 508 to improve performance and responsiveness.

[0126] The SoC 504 can further include a Mobile Industry Processor Interface (MIPI) camera serial interface for receiving video and input from the cameras, a high-speed interface, and / or a video input block that can be used for camera and related pixel input functions. The SoC 504 can further include an input / output controller that can be controlled by software and can be used to receive I / O signals that are not committed to a particular role.

[0127] SoC 504 can further include a wide range of peripheral device interfaces to enable communication with peripherals, audio codecs, power management, and / or other devices. SoC 504 can be used to process data from cameras (connected over Gigabit Multimedia Serial Link and Ethernet), sensors (e.g., LIDAR sensor 564, RADAR sensor 560, etc. that can be connected over Ethernet), data from bus 502 (e.g., speed of vehicle 500, steering wheel position, etc.), data from GNSS sensor 558 (connected over Ethernet or CAN bus). SoC 504 can further include a dedicated high-performance mass storage controller, which can include their own DMA engine, and which can be used to free up CPU 506 from routine data management tasks.

[0128] SoC 504 can be an end-to-end platform with a flexible architecture that spans automation levels 3-5, providing an integrated functional safety architecture for a platform that leverages and efficiently uses computer vision and ADAS technology to achieve diversity and redundancy, along with deep learning tools. SoC 504 can be faster, more reliable, and even more energy and space efficient than conventional systems. For example, accelerators 514, when combined with CPU 506, GPU 508, and data storage 516, can provide a fast and efficient platform for level 3-5 autonomous vehicles.

[0129] The technology thus provides capabilities and functionality that cannot be achieved by conventional systems. For example, computer vision algorithms can be executed on CPUs that can be configured using high-level programming languages such as the C programming language to perform a wide variety of processing algorithms across a wide variety of visual data. However, CPUs often cannot meet the performance requirements of many computer vision applications, such as those related to, for example, execution time and power consumption. In particular, many CPUs cannot execute complex object detection algorithms in real time, which is a requirement for on-board ADAS applications and a requirement for practical level 3-5 autonomous vehicles.

[0130] In contrast to conventional systems, by providing a CPU complex, a GPU complex, and a hardware acceleration cluster, the technology described herein allows multiple neural networks to be executed simultaneously and / or sequentially, and the results to be combined together to achieve level 3-5 autonomous driving functionality. For example, a CNN executed on a DLA or dGPU (e.g., GPU 520) can include text and word recognition, allowing a supercomputer to read and understand traffic signs, including signs for which a neural network has not been specifically trained. The DLA can further include a neural network that is able to recognize, interpret, and provide a semantic understanding of the sign, and pass that semantic understanding to a path planning module running on the CPU complex.

[0131] As another example, multiple neural networks can be run simultaneously as required for level 3, 4, or 5 driving. For example, a warning sign consisting of the words "Flashing lights indicate icy conditions" along with a light can be interpreted by several neural networks independently or collectively. The sign itself can be recognized by a first deployed neural network (e.g., a trained neural network) as a traffic sign, the text "Flashing lights indicate icy conditions" can be interpreted by a second deployed neural network that informs the vehicle's path planning software (preferably executing on the CPU complex) that icy conditions exist when flashing lights are detected. The flashing lights can be recognized by operating a third deployed neural network over multiple frames that informs the vehicle's path planning software of the presence (or absence) of flashing lights. All three neural networks can be run simultaneously, for example, within the DLA and / or on the GPU 508.

[0132] In some examples, a CNN for face recognition and owner recognition can use data from the camera sensors to recognize the presence of an authorized driver and / or owner of the vehicle 500. A processing engine always on the sensors can be used to unlock the vehicle and turn on the lights when the owner approaches the driver's door, and in a safe mode, disable the vehicle when the owner leaves the vehicle. In this way, the SoC 504 provides security against theft and / or carjacking.

[0133] In another example, a CNN for emergency vehicle detection and recognition can use data from the microphones 596 to detect and recognize emergency vehicle sirens. In contrast to conventional systems that detect sirens using a general classifier and manually extract features, the SoC 504 uses a CNN to classify ambient and urban sounds as well as to classify visual data. In a preferred embodiment, a CNN running on the DLA is trained to recognize the relative closing speed of an emergency vehicle (e.g., by using the Doppler effect). The CNN can also be trained to recognize emergency vehicles specific to the local area in which the vehicle is operating as recognized by the GNSS sensor 558. Thus, for example, when operating in Europe, the CNN will seek to detect European sirens, and when in the United States, the CNN will seek to recognize sirens that are only North American. Once an emergency vehicle is detected, a control program can be used to execute an emergency vehicle safety routine that slows the vehicle, pulls over to the side of the road, stops the vehicle, and / or idles the vehicle until the emergency vehicle passes, with the assistance of the ultrasonic sensors 562.

[0134] The vehicle can include a CPU 518 (e.g., a discrete CPU or dCPU) that can be coupled to the SoC 504 via a high-speed interconnect (e.g., PCIe). The CPU 518 can include, for example, an X86 processor. The CPU 518 can be used to perform any of a wide variety of functions, including, for example, arbitrating potentially inconsistent results between ADAS sensors and the SoC 504, and / or monitoring the status and health of the controller 536 and / or infotainment SoC 530.

[0135] The vehicle 500 can include a GPU 520 (e.g., a discrete GPU or dGPU) that can be coupled to the SoC 504 via a high-speed interconnect (e.g., NVIDIA’s NVLINK). The GPU 520 can provide additional artificial intelligence functionality, for example, by executing redundant and / or different neural networks, and can be used to train and / or update neural networks based on input (e.g., sensor data) from sensors of the vehicle 500.

[0136] The vehicle 500 can further include a network interface 524 that can include one or more wireless antennas 526 (e.g., one or more wireless antennas for different communication protocols, such as cellular antennas, Bluetooth antennas, etc.). The network interface 524 can be used to enable wireless connections through the Internet with a cloud (e.g., with the server 578 and / or other network devices), with other vehicles, and / or with computing devices (e.g., client devices of passengers). For communication with other vehicles, a direct link can be established between the two vehicles, and / or an indirect link can be established (e.g., across a network and through the Internet). The direct link can be provided using a car-to-car communication link. The car-to-car communication link can provide the vehicle 500 with information about vehicles that are approaching the vehicle 500 (e.g., vehicles in front of, to the side of, and / or behind the vehicle 500). This functionality can be part of a cooperative adaptive cruise control functionality of the vehicle 500.

[0137] The network interface 524 can include a SoC that provides modulation and demodulation functionality and enables the controller 536 to communicate over a wireless network. The network interface 524 can include a radio frequency front end for up-conversion from baseband to radio frequency and down-conversion from radio frequency to baseband. The frequency conversion can be performed through well-known processes, and / or can be performed using a super-heterodyne process. In some examples, the radio frequency front end functionality can be provided by a separate chip. The network interface can include wireless functionality for communication over LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-Wave, ZigBee, LoRaWAN, and / or other wireless protocols.

[0138] The vehicle 500 can further include a data store 528, which can include off-chip (e.g., off-SoC 504) storage. The data store 528 can include one or more storage elements, including RAM, SRAM, DRAM, VRAM, flash memory, hard disks, and / or other components and / or devices that can store data for at least one bit.

[0139] The vehicle 500 can further include a GNSS sensor 558. The GNSS sensor 558 (e.g., GPS, assisted GPS sensor, differential GPS (DGPS) sensor, etc.) is used to assist in mapping, perception, occupancy grid generation, and / or path planning functions. Any number of GNSS sensors 558 can be used, including, for example and without limitation, a GPS using a USB connector with an Ethernet-to-serial (RS-232) bridge.

[0140] The vehicle 500 can further include a RADAR sensor 560. The RADAR sensor 560 can be used by the vehicle 500 for long-range vehicle detection, even in darkness and / or adverse weather conditions. The RADAR functional safety level can be ASIL B. The RADAR sensor 560 can use the CAN and / or bus 502 (e.g., to transmit data generated by the RADAR sensor 560) for control as well as access to object tracking data, in some examples, Ethernet for access to raw data. A wide variety of RADAR sensor types can be used. For example and without limitation, the RADAR sensor 560 can be suitable for front, rear, and side RADAR use. In some examples, a pulsed Doppler RADAR sensor is used.

[0141] The RADAR sensor 560 can include different configurations, such as long-range with narrow field of view, short-range with wide field of view, short-range side coverage, and so on. In some examples, long-range RADAR can be used for adaptive cruise control functionality. Long-range RADAR systems can provide a wide field of view (e.g., 250 m range) implemented through two or more independent scans. The RADAR sensor 560 can help distinguish between static and moving objects, and can be used by the ADAS system for emergency brake assist and forward collision warning. The long-range RADAR sensor can include a single-station multi-mode RADAR with multiple (e.g., six or more) fixed RADAR antennas, as well as high-speed CAN and FlexRay interfaces. In examples with six antennas, the central four antennas can create focused beam patterns designed to record the surroundings of the vehicle 500 at higher speed with minimal traffic interference from adjacent lanes. The other two antennas can extend the field of view, making it possible to quickly detect vehicles entering or leaving the lane of the vehicle 500.

[0142] As one example, a mid-range RADAR system can include a range of up to 560 m (front) or 80 m (rear) and a field of view of up to 42 degrees (front) or 550 degrees (rear). A short-range RADAR system can include, but is not limited to, RADAR sensors designed to be mounted at both ends of the rear bumper. When mounted at both ends of the rear bumper, such a RADAR sensor system can create two beams that continuously monitor the rear and blind spots to the sides of the vehicle.

[0143] A short-range RADAR system can be used in an ADAS system for blind spot detection and / or lane change assist.

[0144] The vehicle 500 can further include ultrasonic sensors 562. The ultrasonic sensors 562, which can be placed on the front, rear, and / or sides of the vehicle 500, can be used for parking assist and / or to create and update an occupancy grid. A wide variety of ultrasonic sensors 562 can be used, and different ultrasonic sensors 562 can be used for different detection ranges (e.g., 2.5 m, 4 m). The ultrasonic sensors 562 can operate at an ASIL B functional safety level.

[0145] The vehicle 500 can include LIDAR sensors 564. The LIDAR sensors 564 can be used for object and pedestrian detection, emergency braking, collision avoidance, and / or other functions. The LIDAR sensors 564 can be at an ASIL B functional safety level. In some examples, the vehicle 500 can include multiple LIDAR sensors 564 (e.g., two, four, six, etc.) that can use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).

[0146] In some examples, the LIDAR sensors 564 can be capable of providing a list of objects and their distances for a 360-degree field of view. A commercially available LIDAR sensor 564 can have, for example, an advertised range of approximately 500 m, a precision of 2 cm - 3 cm, and support for a 500 Mbps Ethernet connection. In some examples, one or more flush-mounted LIDAR sensors 564 can be used. In such examples, the LIDAR sensors 564 can be implemented as small devices that can be embedded into the front, rear, sides, and / or corners of the vehicle 500. In such examples, the LIDAR sensors 564 can provide a field of view of up to 120 degrees horizontal and 35 degrees vertical, with a range of 200 m, even for low reflectivity objects. Front-mounted LIDAR sensors 564 can be configured for a horizontal field of view between 45 degrees and 135 degrees.

[0147] In some examples, LIDAR technology such as 3D Flash LIDAR can also be used. 3D Flash LIDAR uses a flash of laser light as a source of emission to illuminate the vehicle’s surroundings up to about 200 m. The flash LIDAR unit includes a receptor that records the laser pulse transmission time and reflected light on each pixel, which in turn corresponds to the range from the vehicle to the object. Flash LIDAR can allow for the generation of highly accurate and distortion-free images of the surroundings with each laser flash. In some examples, four flash LIDAR sensors can be deployed, one on each side of the vehicle 500. Available 3D flash LIDAR systems include solid-state 3D staring array LIDAR cameras (e.g., non-scanning LIDAR devices) that have no moving parts other than a fan. The flash LIDAR device can use 5 nanosecond Class I (eye-safe) laser pulses per frame and can capture the reflected laser light in the form of a 3D range point cloud and co-registered intensity data. By using flash LIDAR, and because flash LIDAR is a solid-state device with no moving parts, the LIDAR sensor 564 can be less susceptible to motion blur, vibration, and / or jostling.

[0148] The vehicle can further include an IMU sensor 566. In some examples, the IMU sensor 566 can be located at the center of the rear axle of the vehicle 500. The IMU sensor 566 can include, for example and without limitation, an accelerometer, a magnetometer, a gyroscope, a magnetic compass, and / or other sensor types. In some examples, the IMU sensor 566 can include an accelerometer and a gyroscope, for example in a six-axis application, while in a nine-axis application the IMU sensor 566 can include an accelerometer, a gyroscope, and a magnetometer.

[0149] In some embodiments, the IMU sensor 566 can be implemented as a microelectromechanical systems (MEMS) inertial navigation system (INS) that combines a microelectromechanical systems (MEMS) inertial sensor, a high-sensitivity GPS receiver, and an advanced Kalman filter algorithm to provide estimates of position, velocity, and attitude. As such, in some examples, the IMU sensor 566 can enable the vehicle 500 to estimate heading without input from a magnetic sensor by directly observing the change in velocity from GPS to the IMU sensor 566 and correlating it. In some examples, the IMU sensor 566 and the GNSS sensor 558 can be combined into a single integrated unit.

[0150] The vehicle can include a microphone 596 placed in and / or around the vehicle 500. The microphone 596 can be used for emergency vehicle detection and identification, among other things.

[0151] The vehicle can further include any number of camera types, including stereo cameras 568, wide-view cameras 570, infrared cameras 572, surround-view cameras 574, long and / or mid-range cameras 598, and / or other camera types. These cameras can be used to capture image data around the entire periphery of the vehicle 500. The types of cameras used depend on the embodiment and requirements of the vehicle 500, and any combination of camera types can be used to provide the necessary coverage around the vehicle 500. Further, the number of cameras can vary depending on the embodiment. For example, the vehicle can include six cameras, seven cameras, ten cameras, twelve cameras, and / or another number of cameras. As one example and without limitation, the cameras can support Gigabit Multimedia Serial Link (GMSL) and / or Gigabit Ethernet. Each of the cameras is described in more detail herein with respect to Figure 5A and Figure 5B are described in more detail.

[0152] The vehicle 500 can further include vibration sensors 542. The vibration sensors 542 can measure vibrations of components of the vehicle, such as axles. For example, changes in vibration can indicate changes in the road surface. In another example, when two or more vibration sensors 542 are used, differences between the vibrations can be used to determine the friction or slip of the road surface (e.g., when there is a difference in vibration between a power driven axle and a free spinning axle).

[0153] The vehicle 500 can include an ADAS system 538. In some examples, the ADAS system 538 can include a SoC. The ADAS system 538 can include adaptive / automatic / autonomous cruise control (ACC), cooperative adaptive cruise control (CACC), forward collision warning (FCW), automatic emergency braking (AEB), lane departure warning (LDW), lane keeping assist (LKA), blind spot warning (BSW), rear cross-traffic warning (RCTW), collision warning system (CWS), lane centering (LC), and / or other features and functionality.

[0154] The ACC system can use RADAR sensors 560, LIDAR sensors 564, and / or cameras. The ACC system can include longitudinal ACC and / or lateral ACC. Longitudinal ACC monitors and controls the distance to the vehicle immediately ahead of the vehicle 500 and automatically adjusts the vehicle speed to maintain a safe distance from the vehicle ahead. Lateral ACC performs distance keeping and, if necessary, suggests a lane change for the vehicle 500. Lateral ACC is related to other ADAS applications such as LCA and CWS.

[0155] CACC uses information from other vehicles, which can be received from other vehicles via a wireless link via the network interface 524 and / or wireless antenna 526 or indirectly through a network connection, such as through the Internet. Direct links can be provided by vehicle-to-vehicle (V2V) communication links, while indirect links can be infrastructure-to-vehicle (I2V) communication links. Generally, V2V communication concepts provide information about the immediately preceding vehicles, such as vehicles immediately ahead of and in the same lane as the vehicle 500, while I2V communication concepts provide information about traffic further ahead. A CACC system can include either or both of I2V and V2V information sources. Given information about vehicles ahead of the vehicle 500, CACC can be more reliable, and it has the potential to improve traffic flow and reduce road congestion.

[0156] FCW systems are designed to alert the driver to a hazard so that the driver can take corrective action. FCW systems use a front-facing camera and / or RADAR sensor 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as displays, speakers, and / or vibrating components. FCW systems can provide warnings in the form of, for example, sound, visual warnings, vibrations, and / or quick brake pulses.

[0157] AEB systems detect an impending forward collision with another vehicle or other object and can automatically apply the brakes if the driver does not take corrective action within specified time or distance parameters. AEB systems can use a front-facing camera and / or RADAR sensor 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC. When an AEB system detects a hazard, it typically first alerts the driver to take corrective action to avoid a collision, and if the driver does not take corrective action, the AEB system can automatically apply the brakes in an effort to prevent or at least mitigate the effects of a predicted collision. AEB systems can include technologies such as dynamic brake support and / or crash imminent braking.

[0158] LDW systems provide visual, audible, and / or tactile warnings, such as steering wheel or seat vibrations, to alert the driver when the vehicle 500 is crossing lane markers. The LDW system is not activated when the driver indicates an intentional lane departure by activating a turn signal. LDW systems can use a front-side facing camera coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as displays, speakers, and / or vibrating components.

[0159] An LKA system is a variation of the LDW system. If the vehicle 500 begins to leave the lane, the LKA system provides a steering input or brake to correct the vehicle 500.

[0160] A BSW system detects and warns the driver of vehicles in the car's blind spot. The BSW system can provide visual, audible, and / or tactile alerts to indicate that merging or changing lanes is unsafe. The system can provide additional warnings when the driver uses a turn signal. The BSW system can use rear-side facing cameras and / or RADAR sensors 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.

[0161] A RCTW system can provide visual, audible, and / or tactile notifications when objects are detected outside the range of the rear-facing camera while the vehicle 500 is backing up. Some RCTW systems include AEB to ensure vehicle brakes are applied to avoid a collision. The RCTW system can use one or more rear-facing RADAR sensors 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.

[0162] Conventional ADAS systems can be prone to false positive results, which can annoy and distract the driver, but typically are not catastrophic because the ADAS system alerts the driver and allows the driver to decide whether the safety condition is truly present and act accordingly. However, in an autonomous vehicle 500, in the case of conflicting results, the vehicle 500 itself must decide whether to heed the results from the primary computer or the secondary computer (e.g., the first controller 536 or the second controller 536). For example, in some embodiments, the ADAS system 538 can be a secondary and / or auxiliary computer for providing perception information to a backup computer plausibility module. The backup computer plausibility monitor can run redundant diverse software on hardware components to detect faults in perception and dynamic driving tasks. The output from the ADAS system 538 can be provided to a supervisory MCU. If the outputs from the primary computer and the secondary computer conflict, the supervisory MCU must determine how to reconcile the conflict to ensure safe operation.

[0163] In some examples, the host computer can be configured to provide a confidence score to the supervisory MCU indicating the host computer's confidence in the selected result. If the confidence score exceeds a threshold, then the supervisory MCU can follow the host computer's direction, regardless of whether the secondary computer provides conflicting or inconsistent results. In the event that the confidence score does not satisfy the threshold and in the event that the host computer and the secondary computer indicate different results (e.g., a conflict), the supervisory MCU can arbitrate between the computers to determine the appropriate result.

[0164] The supervisory MCU can be configured to run a neural network that is trained and configured to determine conditions under which the secondary computer provides false alarms based on the output from the host computer and the secondary computer. Thus, the neural network in the supervisory MCU can learn when the output of the secondary computer can be trusted and when it cannot. For example, when the secondary computer is a RADAR-based FCW system, the neural network in the supervisory MCU can learn when the FCW system is identifying metal objects that are not in fact dangerous, such as drain grates or manhole covers that trigger false alarms. Similarly, when the secondary computer is a camera-based LDW system, the neural network in the supervisory MCU can learn to disregard the LDW when a cyclist or pedestrian is present and lane departure is in fact the safest strategy. In embodiments that include a neural network running on the supervisory MCU, the supervisory MCU can include at least one of a DLA or a GPU suitable for running a neural network with associated memory. In preferred embodiments, the supervisory MCU can include and / or be included as a component of the SoC 504.

[0165] In other examples, the ADAS system 538 can include a secondary computer that performs ADAS functions using traditional computer vision rules. As such, the secondary computer can use classic computer vision rules (if-then), and the presence of a neural network in the supervisory MCU can improve reliability, safety, and performance. For example, the diverse implementation and intentional non-identity make the overall system more fault-tolerant, especially with respect to faults caused by software (or software-hardware interface) functions. For example, if there is a software bug or error in the software running on the host computer and the non-identical software code running on the secondary computer provides the same overall result, then the supervisory MCU can be more confident that the overall result is correct and that the bug in the software or hardware on the host computer did not cause a substantial error.

[0166] In some examples, the output of the ADAS system 538 can be fed to a perception block of the host computer and / or a dynamic driving task block of the host computer. For example, if the ADAS system 538 indicates a forward collision warning due to an object immediately ahead, the perception block can use this information in identifying the object. In other examples, the secondary computer can have its own neural network that is trained and thus reduces the risk of false positives as described herein.

[0167] The vehicle 500 can further include an infotainment SoC 530 (e.g., an in-vehicle infotainment system (IVI)). Although illustrated and described as a SoC, the infotainment system can not be a SoC and can include two or more discrete components. The infotainment SoC 530 can include a combination of hardware and software that can be used to provide audio (e.g., music, a personal digital assistant, navigation instructions, news, radio, etc.), video (e.g., TV, movies, streaming media, etc.), telephony (e.g., hands-free calling), network connectivity (e.g., LTE, Wi-Fi, etc.), and / or information services (e.g., a navigation system, a park assist, a telematics device, a radio data system, vehicle-related information such as fuel level, total distance covered, brake fluid level, oil level, doors open / closed, air filter information, etc.) to the vehicle 500. For example, the infotainment SoC 530 can include a radio, a disc player, a navigation system, a video player, USB and Bluetooth connectivity, an in-car computer, in-car entertainment, Wi-Fi, steering wheel audio controls, hands-free voice controls, a heads-up display (HUD), the HMI display 534, a telematics device, a control panel (e.g., for controlling and / or interacting with various components, features, and / or systems), and / or other components. The infotainment SoC 530 can further be used to provide information (e.g., visual and / or audible) to a user of the vehicle, such as information from the ADAS system 538, autonomous driving information such as planned vehicle maneuvers, trajectories, surrounding environment information (e.g., intersection information, vehicle information, road information, etc.), and / or other information.

[0168] The infotainment SoC 530 can include GPU functionality. The infotainment SoC 530 can communicate with other devices, systems, and / or components of the vehicle 500 over the bus 502 (e.g., a CAN bus, Ethernet, etc.). In some examples, the infotainment SoC 530 can be coupled to a supervisory MCU such that, in the event of a failure of the host controller 536 (e.g., a primary and / or backup computer of the vehicle 500), the GPU of the infotainment system can perform some autonomous driving functions. In such examples, the infotainment SoC 530 can place the vehicle 500 in a driver safe park mode as described herein.

[0169] Vehicle 500 may further include an instrument cluster 532 (e.g., a digital instrument panel, electronic instrument cluster, digital instrument panel, etc.). The instrument cluster 532 may include a controller and / or a supercomputer (e.g., a discrete controller or supercomputer). The instrument cluster 532 may include a set of instruments such as a speedometer, fuel level, oil pressure, tachometer, odometer, turn indicator, shift position indicator, seatbelt warning light, parking brake warning light, engine malfunction indicator, airbag (SRS) system information, lighting controls, safety system controls, navigation information, etc. In some examples, information may be displayed and / or shared between the infotainment SoC 530 and the instrument cluster 532. In other words, the instrument cluster 532 may be included as part of the infotainment SoC 530, or vice versa.

[0170] Figure 5D For cloud-based servers and according to some embodiments of this disclosure Figure 5A This is a system diagram illustrating communication between example autonomous vehicles 500. System 576 may include server 578, network 590, and vehicles including vehicle 500. Server 578 may include multiple GPUs 584(A)-584(H) (collectively referred to herein as GPU 584), PCIe switches 582(A)-582(H) (collectively referred to herein as PCIe switch 582), and / or CPUs 580(A)-580(B) (collectively referred to herein as CPU 580). GPUs 584, CPUs 580, and PCIe switches may be interconnected with high-speed interconnects and / or PCIe connections 586, such as, but not limited to, NVLink interfaces 588 developed by NVIDIA. In some examples, GPUs 584 are connected via NVLink and / or NVSwitch SoCs, and GPUs 584 and PCIe switches 582 are connected via PCIe interconnects. Although eight GPUs 584, two CPUs 580, and two PCIe switches are shown in the diagram, this is not intended to be limiting. Depending on the embodiment, each of the servers 578 may include any number of GPUs 584, CPUs 580, and / or PCIe switches. For example, each of the servers 578 may include eight, sixteen, thirty-two, and / or more GPUs 584.

[0171] The server 578 can receive image data from vehicles over the network 590, the image data representing images showing unexpected or changing road conditions such as a road work that has recently started. The server 578 can transmit neural networks 592, updated neural networks 592, and / or map information 594, including information about traffic and road conditions, to vehicles over the network 590. Updates to the map information 594 can include updates to the HD map 522, e.g., information about construction sites, potholes, curves, flooding, or other obstacles. In some examples, the neural networks 592, updated neural networks 592, and / or map information 594 can have been generated from experience using training performed at a data center (e.g., using the server 578 and / or other servers) and / or from data received from any number of vehicles in the environment.

[0172] The server 578 can be used to train machine learning models (e.g., neural networks) based on training data. The training data can be generated by vehicles and / or can be generated in simulations (e.g., using game engines). In some examples, the training data is labeled (e.g., in cases where the neural network benefits from supervised learning) and / or undergoes other pre-processing, while in other examples, the training data is not labeled and / or pre-processed (e.g., in cases where the neural network does not require supervised learning). The training can be performed according to any class or more classes of machine learning techniques, including but not limited to the following classes: supervised training, semi-supervised training, unsupervised training, self-learning, reinforcement learning, federated learning, transfer learning, feature learning (including principal component and cluster analysis), multilinear subspace learning, manifold learning, representation learning (including spare dictionary learning), rule-based machine learning, anomaly detection, and any variants or combinations thereof. Once the machine learning models are trained, the machine learning models can be used by vehicles (e.g., transmitted to vehicles over the network 590), and / or the machine learning models can be used by the server 578 to remotely monitor vehicles.

[0173] In some examples, the server 578 can receive data from vehicles and apply the data to the latest real-time neural networks for real-time intelligent inference. The server 578 can include deep learning supercomputers and / or specialized AI computers powered by GPUs 584, such as the DGX and DGX Station machines developed by NVIDIA. However, in some examples, the server 578 can include deep learning infrastructure of a data center that is powered by CPUs only.

[0174] The deep learning infrastructure of the server 578 can be capable of fast real-time inference, and can use this capability to assess and validate the health of the processors, software, and / or associated hardware in the vehicle 500. For example, the deep learning infrastructure can receive periodic updates from the vehicle 500, such as a sequence of images and / or objects located in the sequence of images that the vehicle 500 has located (e.g., via computer vision and / or other machine learning object classification techniques). The deep learning infrastructure can run its own neural network to identify the objects and compare them to the objects identified by the vehicle 500, and if the results do not match and the infrastructure concludes that the AI in the vehicle 500 is malfunctioning, the server 578 can transmit a signal to the vehicle 500 instructing the fail-safe computer of the vehicle 500 to take control, notify the passengers, and complete a safe parking operation.

[0175] For inference, the server 578 can include GPUs 584 and one or more programmable inference accelerators (such as NVIDIA’s TensorRT 3). The combination of GPU-powered servers and inference-accelerated can make real-time responses possible. In other examples, such as where performance is less important, CPU-, FPGA-, and other processor-powered servers can be used for inference.

[0176] Example Computing Device

[0177] Figure 6 A block diagram of an example computing device 600 suitable for implementing some embodiments of the present disclosure is provided. The computing device 600 can include an interconnect system 602 that directly or indirectly couples the following devices: memory 604, one or more central processing units (CPUs) 606, one or more graphics processing units (GPUs) 608, a communication interface 610, input / output (I / O) ports 612, I / O components 614, a power supply 616, one or more presentation components 618 (e.g., a display), and one or more logic units 620. In at least one embodiment, the computing device 600 can include one or more virtual machines (VMs), and / or any component thereof can include a virtual component (e.g., a virtual hardware component). For a non-limiting example, the one or more GPUs 608 can include one or more vGPUs, the one or more CPUs 606 can include one or more vCPUs, and / or the one or more logic units 620 can include one or more virtual logic units. Thus, the computing device 600 can include discrete components (e.g., a full GPU dedicated to the computing device 600), virtual components (e.g., a portion of a GPU dedicated to the computing device 600), or a combination thereof.

[0178] Although Figure 6various blocks are shown as being connected via the interconnection system 602 having a line, but this is intended to be a simplified representation of a more complex connection that can be present. For example, in some embodiments, a rendering component 618, such as a display device, can be considered an I / O component 614 (e.g., if the display is a touchscreen). As another example, the CPU 606 and / or GPU 608 can include memory (e.g., the memory 604 can represent a storage device in addition to the memory of the GPU 608, CPU 606, and / or other components). In other words, Figure 6 The computing device of FIG. 6 is merely illustrative. Distinctions are not made between “workstation,” “server,” “laptop,” “desktop,” “tablet,” “client device,” “mobile device,” “hand-held device,” “game console,” “electronic control unit (ECU),” “virtual reality system,” and / or other device or system types, as all are contemplated Figure 6 within the scope of the computing device of FIG. 6.

[0179] The interconnection system 602 can represent one or more links or buses, such as an address bus, a data bus, a control bus, or a combination thereof. The interconnection system 602 can include one or more link or bus types, such as an industry standard architecture (ISA) bus, an extended industry standard architecture (EISA) bus, a video electronics standards board (VESA) bus, a peripheral component interconnect (PCI) bus, a peripheral component interconnect express (PCIe) bus, and / or another type of bus or link. In some embodiments, there are direct connections between components. As an example, the CPU 606 can be directly connected to the memory 604. Also, the CPU 606 can be directly connected to the GPU 608. Where there are direct or point-to-point connections between components, the interconnection system 602 can include a PCIe link to perform the connection. In these examples, a PCI bus need not be included in the computing device 600.

[0180] The memory 604 can include any of a wide variety of computer-readable media. Computer-readable media can be any available media that can be accessed by the computing device 600. Computer-readable media can include volatile and nonvolatile media, and removable and non-removable media. By way of example, and not limitation, computer-readable media can comprise computer storage media and communication media.

[0181] Computer storage media can include volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules, and / or other data types. For example, memory 604 can store computer readable instructions (e.g., representing a program and / or program elements, such as an operating system). Computer storage media can include, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by computing device 600. When information is here said to be "stored on" a computer storage medium, such as memory 604, it is meant that the information is stored in memory 604, or in another computer storage medium accessible by computing device 600.

[0182] Computer storage media can include computer readable instructions, data structures, program modules, and / or other data types in a modulated data signal, such as a carrier wave or other transport mechanism, and includes any information delivery media. The term "modulated data signal" can refer to a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, computer storage media can include wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of the any of the above should also be included within the scope of computer readable media.

[0183] CPUs 606 can be configured to execute at least some of the computer readable instructions in order to control one or more components of computing device 600 to perform one or more of the methods and / or processes described herein. Each of CPUs 606 can include one or more cores (e.g., one, two, four, eight, twenty-eight, seventy-two, etc.) capable of processing a large number of software threads simultaneously. CPUs 606 can include any type of processors and can include different types of processors depending on the type of computing device 600 being implemented (e.g., processors with fewer cores for mobile devices and processors with more cores for servers). For example, depending on the type of computing device 600, the processors can be Advanced RISC Machines (ARM) processors implemented using reduced instruction set computing (RISC) or x86 processors implemented using complex instruction set computing (CISC). Computing device 600 can include one or more CPUs 606 in addition to one or more microprocessors or supplemental co-processors such as math co-processors.

[0184] In addition or alternatively to CPU 606, GPU 608 can be configured to execute at least some computer-readable instructions to control one or more components of computing device 600 to perform one or more methods and / or processes described herein. GPU(s) 608 can be integrated GPUs (e.g., with CPU(s) 606) and / or GPU(s) 608 can be discrete GPUs. In embodiments, GPU(s) 608 can be co-processors to CPU(s) 606. Computing device 600 can use GPU(s) 608 to render graphics (e.g., 3D graphics) or perform general-purpose computing. For example, GPU(s) 608 can be used for general-purpose computing on GPUs (GPGPU). GPU(s) 608 can include hundreds or thousands of cores capable of processing hundreds or thousands of software threads concurrently. GPU(s) 608 can generate pixel data for output images in response to rendering commands (e.g., rendering commands from CPU(s) 606 received via a host interface). GPU(s) 608 can include graphics memory, such as display memory, for storing pixel data or any other suitable data (e.g., GPGPU data). Display memory can be included as part of memory 604. GPU(s) 608 can include two or more GPUs operating in parallel (e.g., via a link). The link can connect the GPUs directly (e.g., using NVLINK) or through a switch (e.g., using NVSwitch). When combined together, each GPU 608 can generate pixel data or GPGPU data for different portions of an output or for different outputs (e.g., a first GPU for a first image and a second GPU for a second image). Each GPU can include its own memory or can share memory with other GPUs.

[0185] In addition or alternatively to CPU 606 and / or GPU 608, logic unit(s) 620 can be configured to execute at least some computer-readable instructions to control one or more components of computing device 600 to perform one or more methods and / or processes described herein. In embodiments, CPU(s) 606, GPU(s) 608, and / or logic unit(s) 620 can execute any combination of methods, processes, and / or portions thereof discretely or jointly. Logic unit(s) 620 can be part of and / or integrated with CPU(s) 606 and / or GPU(s) 608, and / or logic unit(s) 620 can be discrete components or otherwise external to CPU(s) 606 and / or GPU(s) 608. In embodiments, logic unit(s) 620 can be processors of CPU(s) 606 and / or GPU(s) 608.

[0186] Examples of logic units 620 include one or more processing cores and / or components thereof, such as tensor cores (TCs), tensor processing units (TPUs), pixel vision cores (PVCs), vision processing units (VPUs), graphics processing clusters (GPCs), texture processing clusters (TPCs), streaming multi-processors (SMs), tree traversal units (TTUs), artificial intelligence accelerators (AIAs), deep learning accelerators (DLAs), arithmetic logic units (ALUs), application-specific integrated circuits (ASICs), floating point units (FPUs), input / output (I / O) elements, peripheral component interconnects (PCIs), or peripheral component interconnect express (PCIe) elements, and the like.

[0187] Communication interface 610 can include one or more receivers, transmitters, and / or transceivers that enable computing device 600 to communicate with other computing devices via electronic communication networks, including wired and / or wireless communications. Communication interface 610 can include components and functionality enabling communication over any of a number of different networks, such as wireless networks (e.g., Wi-Fi, Z-Wave, Bluetooth, Bluetooth LE, ZigBee, etc.), wired networks (e.g., communication over Ethernet or InfiniBand), low power wide area networks (e.g., LoRaWAN, SigFox, etc.), and / or the Internet.

[0188] I / O ports 612 can enable computing device 600 to be logically coupled to other devices including I / O components 614, presentation components 618, and / or other components, some of which can be built into (e.g., integrated with) computing device 600. Illustrative I / O components 614 include a microphone, mouse, keyboard, joystick, game pad, game controller, dish satellite antenna, browser, printer, wireless device, etc. I / O components 614 can provide a natural user interface (NUI) that processes audio, speech, or other physiological inputs generated by a user. In some instances, the inputs can be transmitted to an appropriate network element for further processing. The NUI can implement any combination of speech recognition, stylus recognition, facial recognition, biometric recognition, gesture recognition both on screen and adjacent to the screen, air gestures, head and eye tracking, and touch recognition (as described in more detail below) associated with a display of the computing device 600. Computing device 600 can include depth cameras, infrared cameras, RGB cameras, touch screen technology, and combinations of these, such as a stereoscopic camera system to capture depth camera images.

[0189] Power supply 616 can include a hardwired power supply, a battery power supply, or a combination thereof. Power supply 616 can supply power to computing device 600 to enable operation of components of computing device 600.

[0190] Presentation component 618 can include a display (e.g., a monitor, a touchscreen, a television screen, a heads-up display (HUD), other display types, or a combination thereof), speakers, and / or other presentation components. Presentation component 618 can receive data from other components (e.g., GPU 608, CPU 606, etc.) and output the data (e.g., as images, video, sound, etc.).

[0191] Example data center

[0192] Figure 7 An example data center 700 is shown, which can be used in at least one embodiment of the present disclosure. Data center 700 can include a data center infrastructure layer 710, a framework layer 720, a software layer 730, and an application layer 740.

[0193] As Figure 7 shown, data center infrastructure layer 710 can include a resource orchestrator 712, grouped computing resources 714, and node computing resources (“node C.R.”) 716(1)-716(N), where “N” represents any whole, positive integer. In at least one embodiment, node C.R. 716(1)-716(N) can include, but are not limited to, any number of central processing units (“CPUs”) or other processors (including accelerators, field programmable gate arrays (FPGAs), graphics processors or graphics processing units (GPUs), etc.), memory devices (e.g., dynamic read-only memory), storage devices (e.g., solid state or disk drives), network input / output (“NW I / O”) devices, network switches, virtual machines (“VMs”), power modules, and cooling modules, etc. In some embodiments, one or more of node C.R. 716(1)-716(N) can correspond to a server having one or more of the above-described computing resources. Further, in some embodiments, node C.R. 716(1)-716(N) can include one or more virtual components, such as a vGPU, a vCPU, etc., and / or one or more of node C.R. 716(1)-716(N) can correspond to a virtual machine (VM).

[0194] In at least one embodiment, the grouped computing resources 714 may include individual groups (not shown) of nodes CR716 housed in one or more racks, or a plurality of racks (also not shown) housed in data centers in various geographic locations. Individual groups of nodes CR716 within the grouped computing resources 714 may include computing, networking, memory, or storage resources that can be configured or allocated to support groups of one or more workloads. In at least one embodiment, several nodes CR716, including CPUs, GPUs, and / or other processors, may be grouped within one or more racks to provide computing resources to support one or more workloads. One or more racks may also include any number of power modules, cooling modules, and / or network switches in any combination.

[0195] Resource coordinator 722 may be configured or otherwise control one or more nodes CR716(1)-716(N) and / or grouped computing resources 714. In at least one embodiment, resource coordinator 722 may include a Software Design Infrastructure (“SDI”) management entity for data center 700. The resource coordinator may include hardware, software, or some combination thereof.

[0196] In at least one embodiment, such as Figure 7 As shown, framework layer 720 may include a job scheduler 732, a configuration manager 734, a resource manager 736, and a distributed file system 738. Framework layer 720 may include a framework of software 732 supporting software layer 730 and / or one or more applications 742 of application layer 740. Software 732 or application 742 may respectively include web-based service software or applications, such as service software or applications provided by Amazon Web Services, Google Cloud, and Microsoft Azure. Framework layer 720 may be, but is not limited to, a free and open-source software web application framework, such as Apache Spark, which can utilize distributed file system 738 for large-scale data processing (e.g., "big data"). TM(hereinafter “Spark”). In at least one embodiment, job scheduler 732 can include a Spark driver to facilitate scheduling of workloads supported by various layers of data center 700. In at least one embodiment, configuration manager 734 can be capable of configuring different layers, such as software layer 730 and framework layer 720 including Spark and a distributed file system 738 for supporting large-scale data processing. Resource manager 736 can manage clustered or grouped computing resources mapped to or allocated for supporting distributed file system 738 and job scheduler 732. In at least one embodiment, clustered or grouped computing resources can include grouped computing resources 714 at data center infrastructure layer 710. Resource manager 736 can coordinate with resource orchestrator 712 to manage these mapped or allocated computing resources.

[0197] In at least one embodiment, software 732 included in software layer 730 can include software used by at least portions of node C.R.s 716(1)-716(N), grouped computing resources 714, and / or distributed file system 738 of framework layer 720. One or more types of software can include, but are not limited to, Internet web page search software, email virus scanning software, database software, and streaming video content software.

[0198] In at least one embodiment, one or more application programs 742 included in application layer 740 can include one or more types of application programs used by at least portions of node C.R.s 716(1)-716(N), grouped computing resources 714, and / or distributed file system 738 of framework layer 720. One or more types of application programs can include, but are not limited to, any number of genomics application programs, cognitive computing and machine learning application programs, including training or inferencing software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), and / or other machine learning application programs used in conjunction with one or more embodiments.

[0199] In at least one embodiment, any of configuration manager 734, resource manager 736, and resource orchestrator 712 can implement any number and type of self-modifying actions based on any number and type of data acquired in any technically feasible manner. Self-modifying actions can relieve data center operators of data center 700 from making possibly poor configuration decisions and can avoid underutilized and / or poorly performing portions of a data center.

[0200] Data center 700 may include tools, services, software, or other resources for training one or more machine learning models or using one or more machine learning models to predict or infer information according to one or more embodiments described herein. For example, a machine learning model can be trained by calculating weight parameters based on a neural network architecture using the software and computing resources described above with respect to data center 700. In at least one embodiment, by using weight parameters calculated through one or more training techniques, information can be inferred or predicted using trained machine learning models corresponding to one or more neural networks, such as, but not limited to, those described herein, using the resources described above with respect to data center 700.

[0201] In at least one embodiment, the data center 700 may use a CPU, application-specific integrated circuit (ASIC), GPU, FPGA, and / or other hardware (or corresponding virtual computing resources) to perform training and / or inference. Furthermore, one or more of the aforementioned software and / or hardware resources may be configured as a service to allow a user to train or perform information inference, such as image recognition, speech recognition, or other artificial intelligence services.

[0202] Example network environment

[0203] A network environment suitable for implementing embodiments of this disclosure may include one or more client devices, servers, network-attached storage (NAS), other backend devices, and / or other device types. Client devices, servers, and / or other device types (e.g., each device) may... Figure 6 The implementation is carried out on one or more instances of computing device 600—for example, each device may include similar components, features, and / or functions of computing device 600. Furthermore, in the case of implementing back-end devices (e.g., servers, NAS, etc.), the back-end devices may be included as part of data center 700, examples of which are described herein. Figure 7 To describe in more detail.

[0204] Components of a network environment can communicate with each other via a network, which can be wired, wireless, or both. A network can include multiple networks, or networks within multiple networks. For example, a network can include one or more wide area networks (WANs), one or more local area networks (LANs), one or more public networks (such as the Internet and / or the Public Switched Telephone Network (PSTN)), and / or one or more private networks. In cases where the network includes a wireless telecommunications network, components such as base stations, communication towers, or even access points (and other components) can provide wireless connectivity.

[0205] Compatible network environments can include one or more peer-to-peer network environments (in which case servers can not be included in the network environment), as well as one or more client-server network environments (in which case one or more servers can be included in the network environment). In a peer-to-peer network environment, functionality described herein with respect to servers can be implemented on any number of client devices.

[0206] In at least one embodiment, the network environment can include one or more cloud-based network environments, distributed computing environments, combinations thereof, and the like. A cloud-based network environment can include a framework layer, a job scheduler, a resource manager, and a distributed file system implemented on one or more servers, which can include one or more core network servers and / or edge servers. The framework layer can include a framework for one or more applications of software and / or application layers to support the software layer. The software or applications can include network-based service software or applications, respectively. In embodiments, one or more client devices can use the network-based service software or applications (e.g., by accessing the service software and / or applications via one or more application programming interfaces (APIs)). The framework layer can be, without limitation, a type of free and open-source software web application framework, such as can be used for large-scale data processing (e.g., “big data”) using the distributed file system.

[0207] The cloud-based network environment can provide cloud computing and / or cloud storage that performs any combination of the computing and / or data storage functionality described herein (or one or more portions thereof). Any of these various functionalities can be distributed across multiple locations from central or core servers (e.g., one or more data centers that can be distributed across states, regions, countries, globally, and the like). The core servers can designate at least a portion of the functionality to edge servers if the connection to the user (e.g., client device) is relatively close to the edge servers. The cloud-based network environment can be private (e.g., limited to a single organization), can be public (e.g., available to many organizations), and / or combinations thereof (e.g., a hybrid cloud environment).

[0208] The client devices can include those described herein with respect to Figure 6At least some components, features and functionality of the example computing device 600 are described. By way of example, and not limitation, a client device can embody a personal computer (PC), a laptop computer, a mobile device, a smartphone, a tablet computer, a smartwatch, a wearable computer, a personal digital assistant (PDA), an MP3 player, a virtual reality headset, a global positioning system (GPS) or device, a video player, a video camera, a surveillance device or system, a vehicle, a watercraft, an aircraft, a virtual machine, a drone, a robot, a hand-held communication device, a hospital device, a gaming device or system, an entertainment system, an in-vehicle computer system, an embedded system controller, a remote control, an appliance, a consumer electronic device, a workstation, an edge device, any combination of these described devices, or any other suitable device.

[0209] The present disclosure can be described in the general context of machine-usable instructions or computer code, including computer-executable instructions such as program modules, being executed by a computer or other machine, such as a personal data assistant or other handheld device. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform particular tasks or implement particular abstract data types. The present disclosure can be practiced in a variety of system configurations, including hand-held devices, consumer electronics, general- purpose computers, more specialty computing devices, etc. The present disclosure can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network.

[0210] As used herein, the term “and / or” with respect to two or more elements should be interpreted as only one element or combination of elements. For example, “element A, element B, and / or element C” can include only element A, only element B, only element C, element A and element B, element A and element C, element B and element C, or element A, B, and C. In addition, “at least one of element A or element B” can include at least one of element A, at least one of element B, or at least one of element A and at least one of element B. Further, “at least one of element A and element B” can include at least one of element A, at least one of element B, or at least one of element A and at least one of element B.

[0211] The subject matter of the present disclosure is described with specificity herein to meet statutory requirements. However, the description itself is not intended to limit the scope of this disclosure. Rather, the inventors have contemplated that the claimed subject matter might also be embodied in other ways, to include different steps or combinations of steps similar to the ones described in this document, in conjunction with other present or future technologies. Moreover, although the terms "step" and / or "block" might be used herein to connote different elements of methods employed, the terms should not be interpreted as implying any particular order among or between various steps herein disclosed unless and except when the order of individual steps is explicitly described.

Claims

1. A method for performing a lane change maneuver, comprising: generating a representation of a surrounding environment of a ego vehicle, the representation indicating at least a ego lane, a target lane adjacent to the ego lane, and a location of one or more objects within at least one of the ego lane or the target lane; identifying, based at least in part on the representation of the environment, a plurality of candidate lane change gaps within the target lane for the ego vehicle to maneuver into; selecting, from the plurality of candidate lane change gaps, one of a first lane change gap and a second lane change gap as a target lane change gap based at least in part on evaluating, for the first lane change gap, a plurality of longitudinal velocity profile candidates and evaluating, for the second lane change gap, the plurality of longitudinal velocity profile candidates; selecting, from the plurality of longitudinal velocity profile candidates, a longitudinal velocity profile for the target lane change gap; and performing a lane change in accordance with the selected longitudinal velocity profile and the selected target first lane change gap, wherein selecting, from the plurality of candidate lane change gaps, one of the first lane change gap and the second lane change gap as a target lane change gap comprises: generating a plurality of projections of the plurality of longitudinal velocity profile candidates on a one-dimensional graph representative of the target lane, wherein evaluating, for the first lane change gap, the plurality of longitudinal velocity profile candidates comprises evaluating, for the first lane change gap, the projections of the plurality of longitudinal profile candidates on the one-dimensional graph, and evaluating, for the second lane change gap, the plurality of longitudinal velocity profile candidates comprises evaluating, for the second lane change gap, the projections of the plurality of longitudinal profile candidates on the one-dimensional graph.

2. The method of claim 1, wherein generating the representation of the environment comprises: generating a lane graph representing at least the ego lane and the target lane; detecting the one or more objects based at least in part on sensor data generated using one or more sensors of the ego vehicle; and assigning the one or more objects to at least one of the ego lane or the target lane based at least in part on the lane graph and the detection.

3. The method of claim 1, wherein, selecting the longitudinal velocity profile from the plurality of longitudinal velocity profile candidates comprises reevaluating the plurality of longitudinal velocity profile candidates for the target lane change gap based at least on selecting the target lane change gap.

4. The method of claim 1, further comprising: determining a lateral profile based at least on matching the lateral profile to the longitudinal velocity profile; and generating a trajectory using the longitudinal velocity profile and the lateral profile, wherein performing the lane change is based at least on the trajectory. evaluating a plurality of lateral profile candidates based at least on selecting the longitudinal velocity profile; 5. The method of claim 4, wherein determining the lateral distribution comprises: and selecting the lateral profile from the plurality of lateral profile candidates based at least in part on evaluating the plurality of lateral profile candidates. ​ 6. The method of claim 1, wherein evaluating the plurality of longitudinal speed profile candidates for the first lane change comprises: evaluating the plurality of longitudinal speed profile candidates based on one or more first criteria, the one or more first criteria comprising at least one of: a speed adaptation duration, an acceleration limit, a deceleration limit, or a fast movement limit.

7. The method of claim 1, wherein selecting the longitudinal velocity profile candidate from the plurality of longitudinal velocity profile candidates comprises: selecting the longitudinal speed profile candidate based on one or more second criteria, the one or more second criteria comprising at least one of: a speed adaptation duration, an acceleration limit, a deceleration limit, a fast movement limit, a stop distance overlap, a courtesy distance, a back-off gap count, or a speed loss amount.

8. The method of claim 1, wherein, evaluating the plurality of longitudinal speed profile candidates for the first lane change gap comprises: determining, for the first lane change gap, at least one first score for a first longitudinal speed profile candidate of the plurality of longitudinal speed profile candidates, determining, for the first lane change gap, at least one second score for a second longitudinal speed profile candidate of the plurality of longitudinal speed profile candidates, generating a first cumulative score for the first lane change gap based at least in part on the at least one first score and the at least one second score; and wherein selecting the first lane change gap is based at least on the first cumulative score for the first lane change gap being greater than a second cumulative score for the second lane change gap.

9. The method of claim 1, wherein, one projection of the plurality of projections comprises a one-dimensional longitudinal projection of future positions of the ego vehicle over a period of time according to a corresponding longitudinal speed profile candidate of the plurality of longitudinal speed profile candidates.

10. The method of claim 9, wherein evaluating the plurality of longitudinal velocity profile candidates for the first lane change gap comprises: comparing the one-dimensional longitudinal projection to one or more one-dimensional longitudinal projections of estimated future positions of the one or more objects over the period of time.

11. The method of claim 1, wherein evaluating the plurality of longitudinal speed profile candidates for the first lane change gap comprises: comparing each longitudinal speed profile candidate to a velocity vector field generated based at least in part on at least one position of at least one of the one or more objects; and penalizing the longitudinal speed profile candidate for violating the velocity vector field.

12. A processor comprising: one or more circuits to perform a lane change according to a longitudinal speed profile and selecting one of a first gap and a second gap from a plurality of candidate gaps, the longitudinal speed profile being identified from a plurality of longitudinal speed profile candidates based at least in part on selecting the first gap from the plurality of candidate gaps, the first gap being selected based at least on evaluating the plurality of longitudinal speed profile candidates with respect to the first gap and evaluating the plurality of longitudinal speed profile candidates with respect to a second gap of the plurality of candidate gaps, the selecting including using a representation of an environment to indicate at least a self-lane, a target lane adjacent to the self-lane, and positions of one or more objects within at least one of the self-lane or the target lane, wherein evaluating the plurality of longitudinal speed profile candidates with respect to the first gap includes comparing a one-dimensional longitudinal projection of future positions of an ego vehicle over a period of time to one or more one-dimensional longitudinal projections of estimated future positions of the one or more objects over the period of time.

13. The processor of claim 12, wherein the representation of the environment is generated based at least in part on: generating a lane map representing at least the ego lane and the target lane; detecting the one or more objects based at least in part on sensor data generated using one or more sensors; and assigning the one or more objects to at least one of the ego lane or the target lane based at least in part on the lane map and the detection.

14. The processor of claim 12, wherein the lane change is further performed in accordance with a lateral profile determined based at least in part on evaluating a plurality of lateral profile candidates and selecting the lateral profile from the plurality of lateral profile candidates based at least in part on evaluating the plurality of lateral profile candidates.

15. A system for performing a lane change maneuver, comprising: one or more processing units to perform operations of: generating a representation of a surrounding environment of a ego vehicle, the representation indicating at least a ego lane, a target lane adjacent to the ego lane, and a location of one or more objects within at least one of the ego lane or the target lane; identifying, based at least in part on the representation of the environment, a plurality of candidate lane change gaps within the target lane for the ego vehicle to maneuver into, at least one of the plurality of candidate lane change gaps being delineated at least in part by a location of an object of the one or more objects in the representation of the environment; selecting, from the plurality of candidate lane change gaps, one of a first lane change gap and a second lane change gap as a target lane change gap based at least in part on evaluating a plurality of longitudinal velocity profile candidates for the first lane change gap and the second lane change gap; for the target lane change gap, identifying a longitudinal velocity profile from the plurality of longitudinal velocity profile candidates based at least in part on evaluating the plurality of longitudinal velocity profile candidates in view of one or more criteria; and performing a lane change in accordance with the longitudinal velocity profile and the target lane change gap, wherein selecting, from the plurality of candidate lane change gaps, one of the first lane change gap and the second lane change gap as a target lane change gap comprises: generating a plurality of projections of the plurality of longitudinal velocity profile candidates on a one-dimensional map representative of the target lane, wherein evaluating the plurality of longitudinal velocity profile candidates for the first lane change gap comprises evaluating projections of a plurality of longitudinal profile candidates on the one-dimensional map for the first lane change gap, and evaluating the plurality of longitudinal velocity profile candidates for the second lane change gap comprises evaluating projections of the plurality of longitudinal profile candidates on the one-dimensional map for the second lane change gap.

16. The system of claim 15, wherein the operations further comprise: determining a lateral profile, wherein performing the lane change is further in accordance with the lateral profile.

17. The system of claim 16, wherein determining the lateral distribution comprises: evaluating a plurality of lateral profile candidates; and selecting the lateral profile from the plurality of lateral profile candidates based at least in part on evaluating the plurality of lateral profile candidates.

18. The system of claim 15, wherein the one or more criteria include one or more of a speed-adaptive duration of time, an acceleration limit, a deceleration limit, a fast movement limit, a stop distance overlap, a courtesy distance, a back-off gap count, or an amount of speed loss.

19. The system of claim 15, wherein identifying a longitudinal velocity profile from the plurality of longitudinal velocity profile candidates comprises: responsive to selecting the first lane change gap, re-evaluating at least two of the plurality of longitudinal velocity profile candidates for the first lane change gap.

20. The system of claim 15, wherein the system is included in at least one of a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing deep learning operations; a system implemented using edge devices; a system implemented using robots; a system incorporating one or more virtual machines (VMs); a system implemented at least partially in a data center; or a system implemented at least partially using cloud computing resources.

Images