Abstract: The present disclosure provides a system and a method for jointly controlling one or multiple connected and automated vehicles (CAVs) and one or multiple human-driven vehicles (HDVs) subject to integer constraints for crossing each of multiple intersections on a road. The method comprises collecting digital representation of states of each of the CAVs, HDVs, and traffic signs, solving an optimization problem jointly optimizing traffic flows based on a macroscopic traffic flow model in a centralized traffic controller (CTC) subject to convex relaxation of the integer constraints, solving a multi-variable mixed-integer programming (MIP) problem in each of multiple intersection traffic controllers (ITCs) optimizing a cost function and minimizing tracking errors in traffic flow values of a microscopic traffic flow model with respect to relaxed traffic flow values from the CTC, and transmitting the optimized values of the control commands to the corresponding CAVs and corresponding traffic signs.

Abstract: A vehicle including a judder identification device includes a first sensor group including a plurality of sensors; a second sensor group including a plurality of sensors different from the plurality of sensors of the first sensor group; a processor configured to predict a judder intensity based on information detected by the plurality of sensors of the first sensor group, predict a judder position based on the information detected by the plurality of sensors of the first sensor group and information detected by the plurality of sensors of the second sensor group, and identify a judder position and a judder intensity based on the predicted judder intensity, a preset judder intensity, the predicted judder position, and a preset judder position; and an outputter configured to output the identified judder position and the identified judder intensity.

Abstract: A method of estimating dimensions of vehicle's exterior defects based on reference dimensions comprising analyzing one or more images captured by one or more image sensors deployed to depict a vehicle to identify one or more reference feature relating to the vehicle, obtaining real-world size of the reference feature(s), computing a pixel to real-world size ratio for the image sensor(s) based on the real-world size of the reference feature(s) and a size in pixels of the reference feature(s) in the image(s), analyzing one or more images captured by the image sensor(s) to identify one or more defects in an exterior of the vehicle, computing a size in pixels of one or more dimensions of each defect; and computing a real-world size of the dimension(s) based on the size in pixels of the respective dimension and the pixel to real-world size ratio.

Abstract: A support system for a working vehicle includes a working vehicle including a vehicle body to travel selectively under manual steering or automatic steering, a controller, and a communicator, the controller is configured or programmed to define a planned travel line for and control the automatic steering based on a reference line, the communicator receiving a replacement line for replacing the reference line, and a server to transmit the replacement line to the communicator. The controller includes a status acquirer to acquire a status of a device provided in or on the working vehicle, and is configured or programmed to, in accordance with the status acquired by the status acquirer, switch between a first mode in which the controller replaces the reference line with the replacement line transmitted from the server and a second mode in which the controller does not replace the reference line with the replacement line.

Abstract: This provides locations for mounting controllers and processing components that effectively employ the roof within a frame covered by a cowling so as to avoid exposure to weather and the environment. The roof is also employed to provide a sensor bar that extends across the vehicle width for a distance that does not interfere with normal vehicle function or generate potential overhangs, which can inadvertently engage objects or vehicles. The bar is sufficient in size and shape so as to allow mounting of a plurality of types of sensors on its top surface and/or recessed within front or rear edges. Such sensors can include visual light cameras for machine vison processes and/or LIDAR of various types and cooperage areas/fields of view—some of which can be recessed within a hollow region of the bar. Additional sensors can be mounted on the truck cab and/or chassis, including visual-light cameras and radars.

Abstract: Systems and methods are disclosed for motion forecasting and planning for autonomous vehicles. For example, a plurality of future traffic scenarios are determined by modeling a joint distribution of actor trajectories for a plurality of actors, as opposed to an approach that models actors individually. As another example, a diversity objective is evaluated that rewards sampling of the future traffic scenarios that require distinct reactions from the autonomous vehicle. An estimated probability for the plurality of future traffic scenarios can be determined and used to generate a contingency plan for motion of the autonomous vehicle. The contingency plan can include at least one initial short-term trajectory intended for immediate action of the AV and a plurality of subsequent long-term trajectories associated with the plurality of future traffic scenarios.

Abstract: A method includes: repeatedly performing following steps according to a preset scheduling period: receiving travelling information acquired by a target vehicle; searching, according to a global path of the target vehicle, a topological map for a target directed edge matching the travelling information; determining a right-of-way node sequence of the target vehicle according to the target directed edge and a coverage range of the target vehicle, where the right-of-way node sequence includes multiple right-of-way nodes, and the right-of-way nodes are nodes on the topological map; and determining right-of-way nodes matching the global path in the right-of-way node sequence as target right-of-way nodes in a case that each of the right-of-way nodes in the right-of-way node sequence is in a vacant, and sending the target right-of-way nodes to the target vehicle.

Abstract: A method controls a first vehicle in respect of an oncoming second vehicle. The method determines a turning situation of the first vehicle, in which an expected first trajectory of the first vehicle crosses an expected second trajectory of the second vehicle, and controls the first vehicle in such a way that, during the turning situation, a predetermined distance between the vehicles is maintained. The control includes an influencing of the direction of travel of the first vehicle.

Abstract: A controller for a vehicle capable of autonomous driving. The controller executes detecting an emergency vehicle traveling around the vehicle based on surrounding environment information, and determining how to deal with the emergency vehicle in response to a detection of the emergency vehicle. In the determining how to deal with the emergency vehicle, the controller executes, at least, determining to cause the vehicle to take an avoidance action for the emergency vehicle, and determining to cause the vehicle to take a preliminary action for the avoidance action.

Abstract: A method for controlling a secondary collision avoidance of a vehicle, includes, while an autonomous vehicle is stopped, identifying the positions of another vehicle and an obstacle around the vehicle in advance, and when a side collision occurs by another vehicle, controlling, by an autonomous driving controller, a driving torque and a braking torque of the vehicle to avoid a secondary collision with a vehicle other than another vehicle or the obstacle which may occur after the side collision, minimizing injury to passengers.

Abstract: A control system for controlling movement of a vehicle, is disclosed. The control system is configured to construct a graph having multiple nodes defining states of the vehicle. The multiple nodes comprise an initial node defining initial state and a target node defining target state. Each pair of nodes is connected by an edge defined by collision-free motion primitives where the nodes comprise motion cusps. Multiple nodes connected through edges form a first path. A first number of motion cusps in the first path is determined and the graph is expanded to add new nodes until a termination condition is met on determining that the first number of motion cusps is above a threshold. The expansion of the graph is subjected to a constraint associated with a total number of motion cusps. Further a second path is determined having lesser motion cusps than the first path.

Abstract: A machine learning model is scanned to detect actual or potential threats. The threats can be detected before execution of the machine learning model or during an isolated execution environment. The threat detection may include performing a machine learning file format check, vulnerability check, tamper check, and stenography check. The machine learning model may also be monitored in an isolated environment during an execution or runtime session. After performing a scan, the system can generate a signature based on actual, potential, or absence of detected threats.

Abstract: Real-time event detection is performed on nodes in an environment using position data that is not available to a node in real time but is delayed. A node performs real time event detection by predicting a position of the node based at least in part on delayed position data. The delayed position data is aligned to other sensor data. Aligning the position data may include predicting a position based on dead reckoning and/or a machine learning model. One or more collections of data, each collection including sensor data and predicted position data, is input to a model that performs event detection.

Abstract: Techniques for determining parking spaces and/or parking trajectories for a vehicle based on multistage filtering are discussed herein. In some examples, a vehicle may navigate to a destination within an environment. The vehicle may receive a set of parking locations and perform filtering operations to determine a subset of the parking locations. For example, the vehicle may use a list of parking characteristics (e.g., conditions) to filter the set of parking spaces. Based on determining the subset of parking spaces, the vehicle may generate a set of candidate trajectories to the subset of parking spaces. The vehicle may determine a subset of the candidate trajectories based on filtering the set of candidate trajectories according to various constraints. Further, the vehicle may determine a cost for the subset of candidate trajectories. Based on the costs, the vehicle may determine a candidate trajectory for the vehicle to follow.

Abstract: Provided is an autonomous travel system capable of effectively suppressing generation of ruts. It is a further object to provide an autonomous travel system including unmanned vehicles that travel on a transportation path constituted of opposite lanes, which is capable of suppressing generation of ruts while preventing proximity to an on-corning vehicle. An in-vehicle control device 200 includes: an offset amount determination unit 202 adapted to, based on common offset information received via a wireless communication device 240, determine an offset amount of a travel path 60 based on map information 251 and generate a target track 62; and an autonomous travel control unit 201 adapted to output a travel instruction to control traveling of a body so as to track the target track 62 to which the offset amount has been added based on the target track 62 and an own-vehicle position.

Abstract: Aspects of the disclosure provide controlling an autonomous vehicle. For example, a first trajectory generated by a first planning system may be received. A portion of the first trajectory may be input into a second planning system. A second trajectory generated by the second planning system based on the portion of the first trajectory may be received. The second trajectory may be used to determine whether to validate the first trajectory, and when the second trajectory is determined to be validated, the autonomous vehicle may be enabled to be controlled based on the first trajectory.

Abstract: Disclosed are computer systems and techniques for authenticating a passenger of an autonomous vehicle and operating doors of the autonomous vehicle. For passenger authentication, the computer system is configured to receive a ride request, generate a passcode, transmit the passcode to a user account and the autonomous vehicle, authenticate the user using the passcode, and enable departure of the autonomous vehicle. For door operation, the computer system is configured to detect environmental conditions surrounding an autonomous vehicle, determine based on a set of operational conditions whether one or more doors of the autonomous vehicle are desirable for use, and operate a door if the door is safe to operate and desirable for operation.

Abstract: A vehicle control device for controlling steering and braking of a vehicle: determines, based on prediction of movement of an object detected on a road and prediction of a traveling trajectory of the vehicle, whether collision between the vehicle and the object will occur; determines whether it is possible to avoid the collision by braking without steering when it is determined that the collision will occur; and performs braking control for avoiding the collision and assists the steering such that the vehicle remains in a lane in which the vehicle is traveling when it is determined that it is possible to avoid the collision by braking.

Abstract: A travel assistance method and device, when performing a first lane change with autonomous lane change control and then performing a second lane change for traveling along a set travel route, determines whether or not a distance from a position at which the first lane change has concluded to a position at which the second lane change can be started is a predetermined distance or less. When it is determined that the distance is the predetermined distance or less, whether assistance with the autonomous lane change control is enabled for the second lane change is determined. When assistance with the autonomous lane change control is enabled, whether a lane change can be made from a subject vehicle lane to an adjacent lane is determined. When the lane change to the adjacent lane can be made, the second lane change is made with the autonomous lane change control.

Abstract: Methods and apparatuses associated with updating a map using images are described. An apparatus can include a processing resource and a memory resource having instructions executable to a processing resource to monitor a map including a plurality of locations, receive, at the processing resource, the memory resource, or both, and from a first source, image data associated with a first location, identify the image data as being associated with a missing portion, an outdated portion, or both, of the map, and update the missing portion, the outdated portion, or both, of the map with the image data.

Abstract: Aspects of the disclosure provide for a method of controlling an autonomous vehicle in an autonomous driving mode. For instance, a predicted future trajectory for an object detected in a driving environment of the autonomous vehicle may be received. A routing intent for a planned trajectory for the autonomous vehicle may be received. The predicted future trajectory and the routing intent intersect with one another may be determined. When the predicted future trajectory and the routing intent are determined to intersect with one another, a time gap may be applied to a predicted future state of the object defined in the predicted future trajectory. A planned trajectory may be determined for the autonomous vehicle based on the applied time gap. The autonomous vehicle may be controlled in the autonomous driving mode based on the planned trajectory.

Abstract: Real-time event detection is performed on nodes in an environment using position data that is not available to a node in real time but is delayed. A node performs real time event detection by predicting a position of the node based at least in part on delayed position data. The delayed position data is aligned to other sensor data. Aligning the position data may include predicting a position based on dead reckoning and/or a machine learning model. One or more collections of data, each collection including sensor data and predicted position data, is input to a model that performs event detection.

Abstract: An information processing device of: generating first routes satisfying a first condition; calculating a first index of each node based on a cost of each edge; generating a second route arriving at each node from a starting point and satisfying a second condition; generating a third route satisfying the second condition by adding the edge to the second route; calculating a second index being a difference between a total value of the first index and the cost of the second and third routes; updating the second route when the second index of the third route is smaller than the second index of the second route; excluding the edge having not contributed to the updating more than a predetermined number of times; and outputting a route with the smallest second index, the first index representing a degree of reduction of the cost in a linearly relaxed problem of the routing problem.

Abstract: A network apparatus, for example, obtains probe data corresponding to a respective vehicle making a late lane change while traversing a TME of a traversable network. Location information indicating a location of the late lane change is extracted from the probe data. Map, weather, and/or traffic data for the location is obtained. A late lane change feature description is generated based on information extracted from the probe data and the map, weather, and/or traffic data. A model is trained using a machine learning technique and training data comprising the late lane change feature description. The model is executed to generate a late lane change prediction corresponding to a TME of a digital map. The network apparatus causes at least one of (a) the digital map to be updated, (b) traffic data corresponding to the TME to be updated, or (c) a navigation-related function to be performed based on the prediction.

Abstract: An ADS performs processing including setting an acceleration command to V1 when an autonomous state is set to an autonomous mode, when the acceleration command has a value indicating deceleration, when a vehicle velocity is zero, and when a standstill command is set to “Applied” and setting an immobilization command to “Applied” when a traveling direction of a vehicle indicates a standstill status, when there is a wheellock request, and when predetermined time has elapsed since the vehicle came to a standstill.

Abstract: System, methods, and computer-readable media for configuring an autonomous vehicle based on safety scores determined by a safety score prediction algorithm, and an associated training technique, to output a safety score for a predicted and/or actual path. The safety score is a value indicating a likelihood of risky events per distance. The safety score prediction algorithm is trained with historical human driving datasets associated with paths (predicted or actual) taken by one or more AVs during human driving.

Abstract: An example method includes (a) obtaining sensor data descriptive of an environment of an autonomous vehicle; (b) obtaining a plurality of travel way markers from map data descriptive of the environment; (c) determining, using a machine-learned object detection model and based on the sensor data, an association between one or more travel way markers of the plurality of travel way markers and an object in the environment; and (d) generating, using the machine-learned object detection model, an offset with respect to the one or more travel way markers of a spatial region of the environment associated with the object.

Abstract: A plurality of vehicle charging stations and a plurality of stationary sensors with respective fields of view can be disposed within an area. A stationary node computer including a first processor and a first memory and communicatively can be coupled to the charging station and the stationary sensors, the first memory storing instructions executable by the first processor such that the stationary computer is programmed to authorize a request from a vehicle to access the charging station; identify an object, detected in the area in data from one or more of the stationary sensors, to be tracked with respect to the vehicle; wherein the object to be tracked with respect to the vehicle is identified based on a type of the object, an attribute of the object and an attribute of the vehicle; and transmit data about the object to the vehicle.

Abstract: A vehicle control system (500) controls a vehicle whereon a plurality of ECUs (30) and a vehicle integrated control device (10) to control the plurality of ECUs (30) are mounted. The vehicle integrated control device (10) includes a control target value operation unit to calculate a control target value to control the plurality of ECUs (30). Further, the vehicle integrated control device (10) includes a prediction control value operation unit to estimate a state of the vehicle in the future, and to calculate a prediction control value to control the plurality of ECUs (30). The vehicle integrated control device (10) includes an instruction signal generation unit to generate an instruction signal including an operation instruction and a prediction control instruction. Each of the plurality of ECUs (30) includes an actuator control unit to control an actuator (50) based on the prediction control instruction.

Abstract: Methods and systems for jointly estimating a pose and a shape of an object perceived by an autonomous vehicle are described. The system includes data and program code collectively defining a neural network which has been trained to jointly estimate a pose and a shape of a plurality of objects from incomplete point cloud data. The neural network includes a trained shared encoder neural network, a trained pose decoder neural network, and a trained shape decoder neural network. The method includes receiving an incomplete point cloud representation of an object, inputting the point cloud data into the trained shared encoder, outputting a code representative of the point cloud data. The method also includes generating an estimated pose and shape of the object based on the code. The pose includes at least a heading or a translation and the shape includes a denser point cloud representation of the object.

Abstract: An embodiment is an autonomous driving system in a heterogeneous SD map and precise map environment, which can provide an autonomous driving service using a heterogeneous SD map having properties that do not match those of a precise map. The autonomous driving system in a heterogeneous SD map and HD map environment includes a service platform and an autonomous vehicle, and the operation and operation method of the system depend on whether a user can input TBT in addition to a start point and a destination.

Abstract: Systems and techniques are provided for providing a preorder delivery in an autonomous vehicle (AV). An example method includes receiving, from a computing device associated with a user, a first user request for a ride in an AV from a user location to a user destination; receiving, from the computing device associated with the user, a second user request for the AV to pick up a product at a pick up location associated with the product, prior to picking up the user; determining, based on the first and second user requests, a route for the AV to navigate; and sending, from a computer associated with the AV to a locker of the AV, an instruction to open the locker where the product is placed.

Abstract: This document describes techniques for localizing a vehicle to map data, for example, to enable assisted driving and automated driving within a road. A road model builder is described that uses vehicle location data to determine its position relative to map data obtained from a provider, which can be different than the location data's source. Using the vehicle position, the road model builder maintains a current lane segment for the vehicle and its relative position within the current lane segment. Based on its relative position, the road model builder determines whether a transition to another lane segment is appropriate before setting the current lane segment to be the other lane segment. The current lane segment, when provided as an input to an assisted-driving or automated-driving system, can enable safer operation of the vehicle and use fewer computing resources than other systems.

Abstract: An apparatus for controlling an autonomous vehicle disclosure may include a processor and a memory configured to be operatively connected to the processor and to store at least one code performed in the processor, wherein the memory may store a code that, when executed by the processor, causes the processor to control the autonomous vehicle to travel on the basis of a distance from a preceding vehicle in a travel lane in which the autonomous vehicle travels or a preset speed, determine a risk level of a lane change on the basis of a speed of the autonomous vehicle, a speed of a side vehicle traveling in a target lane of a lane change, and a distance between the autonomous vehicle and the side vehicle upon occurrence of a lane change request, and perform longitudinal or lateral control for the lane change on the basis of the risk level.

Abstract: A system and a method for controlling movement of a robotic vehicle. The robotic vehicle comprises a computer system, a chassis controller, and a plurality of sensors. Data measured by the plurality of sensors is received by the computer system. A base trajectory and an emergency trajectory for the robotic vehicle are generated based on the data. During normal operations of the robotic vehicle the chassis controller causes the robotic vehicle to travel along the base trajectory. After determining that an error has occurred, the chassis controller causes the robotic vehicle to travel along the emergency trajectory.

Abstract: A number of variations may include a method or system using human driving characteristics in autonomous vehicle driving method or system.

Abstract: This application provides a method for controlling vehicles driving as a platoon performed by an electronic device. The method includes: transmitting a first vehicle clearing message to a target in-platoon vehicle, and a second vehicle clearing message to following vehicles behind the target in-platoon vehicle, the first vehicle clearing message used for instructing the target in-platoon vehicle to leave the platoon, the second vehicle clearing message used for instructing temporary clearing of the following vehicles behind the target in-platoon vehicle; receiving a platoon joining message returned from the following vehicles behind the target in-platoon vehicle; and reorganizing the following vehicles behind the target in-platoon vehicle according to the platoon joining message to form a new vehicle platoon, the new vehicle platoon not including the target in-platoon vehicle and following vehicles having not transmitted the platoon joining message to the leading vehicle.

Abstract: Provided is a vehicle traveling assist device capable of suppressing deterioration of ride comfort of an own vehicle due to vehicle traveling assist. A target determination unit calculates a target speed and a target distance based on target vehicle speed information. A plan generation unit generates a distance plan and a speed plan based on the target speed and the target distance through use of a three-stage moving average filter. The three-stage moving average filter is a filter including a first moving average filter, a second moving average filter, and a third moving average filter which are sequentially and serially connected to each other. A vehicle control unit generates an acceleration command for an own vehicle based on the distance plan and the speed plan.

Abstract: An autonomous mobile robot uses a capability-aware pathfinding algorithm to traverse from a start pose to an end pose efficiently and effectively. The robot receives a start pose and an end pose, and determines a primary path from the start pose to the end pose based on a primary pathfinding algorithm. The robot may smooth the primary path using Bezier curves. The robot may identify a conflict point on the primary path where the robot cannot traverse, and may determine a secondary path from a first point before the conflict point to a second point after the conflict point. The secondary path may use a secondary pathfinding algorithm that uses motion primitives of the robot to generate the secondary path based on the motion capabilities of the robot. The robot may then traverse from the start pose to the end pose based on the primary path and the secondary path.

Abstract: Dynamically adjusting, using artificial intelligence (AI), sensors and models of an autonomous roaming robotic device, which includes receiving data regarding an asset at a computer of a roaming robotic device from sensors on the robotic device. The robotic device identifies an asset at a location using the sensors, and the robotic device has instructions, received from a control system, to inspect the location or items at the location. The data is analyzed using the computer of the robotic device, and the analysis includes using historical data for the asset. An AI model is loaded using the computer of the robotic device, based on the identification of the asset. A sensor is selected using the computer of the robotic device, for conducting an inspection of the asset based on the analysis of the data and the AI model.

Abstract: Techniques for determining a vehicle trajectory that causes a vehicle to navigate in an environment relative to one or more objects are described herein. For example, the techniques may include a computing device determining a decision tree having nodes to represent different object intents and/or nodes to represent vehicle actions at a future time. A tree search algorithm can search the decision tree to evaluate potential interactions between the vehicle and the one or more objects over a time period, and output a vehicle trajectory for the vehicle. The vehicle trajectory can be sent to a vehicle computing device for consideration during vehicle planning, which may include simulation.

Abstract: Systems and methods are provided that may to cause autonomous navigation of autonomous vehicles in the case of non-standard traffic flows such as police stops, emergency vehicle passing, construction sites, vehicle collision sites, and other non-standard road conditions. An entity associated with the non-standard traffic flow (e.g., an emergency vehicle, road sign, barrier, etc.) may transmit or broadcast a control signal to be received (or otherwise detected) at one or more autonomous vehicles. Each autonomous vehicle, upon receiving the control signal, may autonomously navigate in accordance with the control signal, thus mitigating or eliminating dangers associated with non-standard traffic flows.

Abstract: The present application discloses a method, system, and computer system for providing a unified map and performing an active measure for re-routing transport based on one or more hazards along a current route. The method includes obtaining context information for a managed vehicle, obtaining map data, determining, based at least in part on the context information and the map data, whether the managed vehicle is parked in an unsafe location, in response to determining that the managed vehicle is parked in the unsafe location, determining whether to perform an active measure with respect to the managed vehicle, and in response to determining to perform the active measure, performing the active measure.

Abstract: The present invention provides an object recognition system for recognising objects within an environment, including: tags, each tag being associated with a respective object, each tag including: a tag memory to store an object model being a computational model indicative of a relationship between the respective object and a shape encoding layer; a sensing system including: a sensing device to sense the environment; a sensing system memory to store a sensing device model being a computational model indicative of a relationship between sensor data captured using the sensing device and the shape encoding layer; sensing system processing devices to: receive object models from a transceiver; retrieve the sensing device model from the sensing system memory; acquire sensor data indicative of the environment from the sensing device; and analyse the sensor data using the sensing device model and the object models to thereby recognise the objects within the environment.

Abstract: Systems and methods for process tending with a robot arm are presented. The system comprises a robot arm and robot arm control system mounted on a self-driving vehicle, and a server in communication with the vehicle and/or robot arm control system. The vehicle has a vehicle control system for storing a map and receiving a waypoint based on a process location provided by the server. The robot arm control system stores at programs that is executable by the robot arm. The vehicle control system autonomously navigates the vehicle to the waypoint based on the map, and the robot arm control system selects a target program from the stored programs based on the process location and/or a process identifier.

Abstract: Disclosed is a vehicle path control apparatus including a memory provided to store danger driving vehicle data, a communication device provided to receive road information from an external server, and a processor configured to identify a danger driving vehicle driving on a road based on the road information and the danger driving vehicle data, select an unmanned vehicle driving within a preset distance from the danger driving vehicle based on the road information, generate a driving path of the unmanned vehicle for obstructing a course of the danger driving vehicle based on an expected path of the danger driving vehicle, and control the communication device to transmit information on the driving path to the unmanned vehicle.

Abstract: A mobile turf sprayer operates to spray a product onto a turf site. In some embodiments the mobile turf sprayer identifies a plurality of regions of the turf site to be sprayed and automatically sprays product onto those regions as the mobile turf sprayer moves about the turf site.

Abstract: Techniques for accurately predicting vehicle state errors for avoiding collisions with objects detected in an environment of a vehicle are discussed herein. A vehicle safety system can implement a model determine a representation of the vehicle usable in a scenario. The model may dynamically determine a size and/or a heading of the vehicle representation based on differences between a candidate trajectory and a current trajectory of the vehicle. The safety system can identify a potential collision between the vehicle and the object based at least in part on an overlap of the vehicle representation and an object representation.

Abstract: Example data processing methods and apparatus are provided. One example method includes obtaining an image captured by an in-vehicle camera. A to-be-detected target in the image is determined. A feature region corresponding to the to-be-detected target in the image is further determined based on a location of the to-be-detected target in the image. A first parking state is determined based on the image and wheel speedometer information. A first homography matrix corresponding to the first parking state is determined from a prestored homography matrix set, where different parking states correspond to different homography matrices. Image information of the feature region is processed based on the first homography matrix to obtain a detection result.

Abstract: Techniques for determining unified futures of objects in an environment are discussed herein. Techniques may include determining a first feature associated with an object in an environment and a second feature associated with the environment and based on a position of the object in the environment, updating a graph neural network (GNN) to encode the first feature and second feature into a graph node representing the object and encode relative positions of additional objects in the environment into one or more edges attached to the node. The GNN may be decoded to determine a first predicted position of the object. The first predicted position may be determined to be outside of a bounded area of the environment. Based on this determination, a second predicted position of the object may be determined using map data associated with the object.