Abstract: Systems and methods for providing a vehicle service are provided. In one example embodiment, a computer-implemented method includes receiving data indicative of a service request to provide a vehicle service for an entity with respect to one or more cargo items designated for autonomous transport. The method includes obtaining a first cargo item among the one or more cargo items, from a representative of the entity at a dedicated first transfer hub proximate to a first location associated with the first cargo item. The method includes controlling a first autonomous vehicle to transport the first cargo item from the first transfer hub to a dedicated second transfer hub proximate to a second location associated with the first cargo item. The method includes providing the first cargo item to a representative of the entity at the second transfer hub, to provide the vehicle service.

Abstract: Aspects of the present disclosure involve systems, methods, and devices for mitigating Lidar cross-talk. Consistent with some embodiments, a method includes detecting a noise signal producing noise in one or more return signals being received by a Lidar unit of an autonomous vehicle (AV) system, and detecting a noise source corresponding to the noise signal. The detecting of the noise source comprises determining a direction of the noise source relative to the AV system and determining a classification of the noise source based on an intensity of the noise signal. The method further includes generating state data to describe the noise source based on the direction of the noise source relative to AV system and the classification of the noise source. The method further includes controlling one or more operations of the AV system based on the state data describing the noise source.

Abstract: Systems and methods for vehicle message signing are provided. A method includes obtaining, by a vehicle computing system of an autonomous vehicle, a computing system state associated with the vehicle computing system and a message from at least one remote process running a computing device remote from the vehicle computing system. The message is associated with an intended recipient process running on the vehicle computing system. The method includes determining an originating sender for the message. The originating sender is indicative of a remote process that generated the message. The method includes determining a routing action for the message based on a comparison of the originating sender and the computing system state. The routing action includes at least one of a discarding action or a forwarding action to the intended recipient process. The method includes performing the routing action for the message.

Abstract: Systems and methods for controlling an autonomous vehicle are provided. In one example embodiment, a computer-implemented method includes obtaining, from an autonomy system, data indicative of a planned trajectory of the autonomous vehicle through a surrounding environment. The method includes determining a region of interest in the surrounding environment based at least in part on the planned trajectory. The method includes controlling one or more first sensors to obtain data indicative of the region of interest. The method includes identifying one or more objects in the region of interest, based at least in part on the data obtained by the one or more first sensors. The method includes controlling the autonomous vehicle based at least in part on the one or more objects identified in the region of interest.

Abstract: Systems, devices, products, apparatuses, and/or methods for generating a road edge boundary for an edge of a road in an AV map for controlling an autonomous vehicle on a roadway by obtaining map data associated with a map of a geographic location including a roadway associated with one or more locations of one or more vehicles in the roadway during one or more traversals of the roadway, determining one or more prediction scores based on the map data, including one or more predictions of whether the plurality of elements include road edge boundary locations, and generating in the map a road edge boundary based on the one or more prediction scores.

Abstract: A computing system can input first relative location embedding data into an interaction transformer model and receive, as an output of the interaction transformer model, motion forecast data for actors relative to a vehicle. The computing system can input the motion forecast data into a prediction model to receive respective trajectories for the actors for a current time step and respective projected trajectories for the actors for a subsequent time step. The computing system can generate second relative location embedding data based on the respective projected trajectories from the second time step. The computing system can produce second motion forecast data using the interaction transformer model based on the second relative location embedding. The computing system can determine second respective trajectories for the actors using the prediction model based on the second forecast data.

Abstract: Various examples are directed to systems and methods for routing an autonomous vehicle. A vehicle autonomy system may generate first route data describing a first route for the autonomous vehicle to a first target location and control the autonomous vehicle using the first route data. The vehicle autonomy system may determine that the autonomous vehicle is within a threshold of the first target location and select a second target location associated with at least a second stopping location. The vehicle autonomy system may generate second route data describing a route extension of the first route from the first target location to the second target location and control the autonomous vehicle using the second route data.

Abstract: A method for determining and providing alternative routes receives request data associated with a request from a user device. A first canonical route is determined from a plurality of canonical routes based on the request data. Each respective canonical route of the plurality of canonical routes satisfies at least one autonomy criteria associated with whether an autonomous vehicle can travel on the respective canonical route. First route data associated with the first canonical route is provided. Route identification data associated with identifying an alternative canonical route is received after providing the first route data associated with the first canonical route. A second canonical route is determined from the plurality of canonical routes based on the route identification data. Second route data associated with the second canonical route is provided for controlling travel of the autonomous vehicle on the second canonical route.

Abstract: An autonomous vehicle which includes multiple independent control systems that provide redundancy as to specific and critical safety situations which may be encountered when the autonomous vehicle is in operation.

Abstract: Generally, the disclosed systems and methods implement improved detection of objects in three-dimensional (3D) space. More particularly, an improved 3D object detection system can exploit continuous fusion of multiple sensors and/or integrated geographic prior map data to enhance effectiveness and robustness of object detection in applications such as autonomous driving. In some implementations, geographic prior data (e.g., geometric ground and/or semantic road features) can be exploited to enhance three-dimensional object detection for autonomous vehicle applications. In some implementations, object detection systems and methods can be improved based on dynamic utilization of multiple sensor modalities. More particularly, an improved 3D object detection system can exploit both LIDAR systems and cameras to perform very accurate localization of objects within three-dimensional space relative to an autonomous vehicle.

Abstract: Systems and methods of the present disclosure provide an improved approach for open-set instance segmentation by identifying both known and unknown instances in an environment. For example, a method can include receiving sensor point cloud input data including a plurality of three-dimensional points. The method can include determining a feature embedding and at least one of an instance embedding, class embedding, and/or background embedding for each of the plurality of three-dimensional points. The method can include determining a first subset of points associated with one or more known instances within the environment based on the class embedding and the background embedding associated with each point in the plurality of points. The method can include determining a second subset of points associated with one or more unknown instances within the environment based on the first subset of points. The method can include segmenting the input data into known and unknown instances.

Abstract: The present disclosure is directed to controlling state transitions in an autonomous vehicle. In particular, a computing system can initiate the autonomous vehicle into a no-authorization state upon startup. The computing system can receive an authorization request. The computing system determines whether the authorization request includes a request to enter the first or second mode of operations, wherein the first mode of operations is associated with the autonomous vehicle being operated without a human operator and the second mode of operations is associated with the autonomous vehicle being operable by a human operator. The computing system can transition the autonomous vehicle from the no-authorization state into a standby state in response to determining the authorization request includes a request to enter the first mode of operations or into a manual-controlled state in response to determining the authorization request is a request to enter the second mode of operations.

Abstract: Systems and methods for detecting objects and predicting their motion are provided. In particular, a computing system can obtain a plurality of sensor sweeps. The computing system can determine movement data associated with movement of the autonomous vehicle. For each sensor sweep, the computing system can generate an image associated with the sensor sweep. The computing system can extract, using the respective image as input to one or more machine-learned models, feature data from the respective image. The computing system can transform the feature data into a coordinate frame associated with a next time step. The computing system can generate a fused image. The computing system can generate a final fused image. The computing system can predict, based, at least in part, on the final fused representation of the plurality of sensors sweeps from the plurality of sensor sweeps, movement associated with the feature data at one or more time steps in the future.

Abstract: Vehicles according to at least some embodiments of the disclosure include a sensor, and a computing device comprising at least one hardware processing unit. The computing device is programmed to perform operations comprising capturing an image with the sensor, identifying an object in the image, and in response to an accuracy of the identification meeting a first criterion, pre-loading a braking system of the autonomous vehicle. In some aspects, the computing device may predict that an object not currently within a path of the vehicle has a probability of entering the path of the vehicle that meets a second criterion. When the probability of entering the path meets the second criterion, some of the disclosed embodiments may pre-load the braking system.

Abstract: Systems and methods for training machine-learned models are provided. A method can include receiving a rasterized image associated with a training object and generating a predicted trajectory of the training object by inputting the rasterized image into a first machine-learned model. The method can include converting the predicted trajectory into a rasterized trajectory that spatially corresponds to the rasterized image. The method can include utilizing a second machine-learned model to determine an accuracy of the predicted trajectory based on the rasterized trajectory. The method can include determining an overall loss for the first machine-learned model based on the accuracy of the predictive trajectory as determined by the second machine-learned model. The method can include training the first machine-learned model by minimizing the overall loss for the first machine-learned model.

Abstract: Systems and methods for vehicle-to-vehicle communications are provided. An example computer-implemented method includes obtaining from a first autonomous vehicle, by a computing system onboard a second autonomous vehicle, a first compressed intermediate environmental representation. The first compressed intermediate environmental representation is indicative of at least a portion of an environment of the second autonomous vehicle and is based at least in part on sensor data acquired by the first autonomous vehicle at a first time. The method includes generating, by the computing system, a first decompressed intermediate environmental representation by decompressing the first compressed intermediate environmental representation. The method includes determining, by the computing system, a first time-corrected intermediate environmental representation based at least in part on the first decompressed intermediate environmental representation.

Abstract: Systems and methods for vehicle-to-vehicle communications are provided. An example computer-implemented method includes obtaining, by a computing system onboard a first autonomous vehicle, sensor data associated with an environment of the first autonomous vehicle. The method includes determining, by the computing system, an intermediate environmental representation of at least a portion of the environment of the first autonomous vehicle based at least in part on the sensor data. The method includes generating, by the computing system, a compressed intermediate environmental representation by compressing the intermediate environmental representation of at least the portion of the environment of the first autonomous vehicle. The method includes communicating, by the computing system, the compressed intermediate environmental representation to a second autonomous vehicle.

Abstract: Generally, the present disclosure is directed to systems and methods for streaming processing within one or more systems of an autonomy computing system. When an update for a particular object or region of interest is received by a given system, the system can control transmission of data associated with the update as well as a determination of other aspects by the given system. For example, the system can determine based on a received update for a particular aspect and a priority classification and/or interaction classification determined for that aspect whether data associated with the update should be transmitted to a subsequent system before waiting for other updates to arrive.

Abstract: Systems and methods for generating motion plans for autonomous vehicles are provided. An autonomous vehicle can include a machine-learned motion planning system including one or more machine-learned models configured to generate target trajectories for the autonomous vehicle. The model(s) include a behavioral planning stage configured to receive situational data based at least in part on the one or more outputs of the set of sensors and to generate behavioral planning data based at least in part on the situational data and a unified cost function. The model(s) includes a trajectory planning stage configured to receive the behavioral planning data from the behavioral planning stage and to generate target trajectory data for the autonomous vehicle based at least in part on the behavioral planning data and the unified cost function.

Abstract: Systems and methods for generating motion plans including target trajectories for autonomous vehicles are provided. An autonomous vehicle may include or access a machine-learned motion planning model including a backbone network configured to generate a cost volume including data indicative of a cost associated with future locations of the autonomous vehicle. The cost volume can be generated from raw sensor data as part of motion planning for the autonomous vehicle. The backbone network can generate intermediate representations associated with object detections and objection predictions. The motion planning model can include a trajectory generator configured to evaluate one or more potential trajectories for the autonomous vehicle and to select a target trajectory based at least in part on the cost volume generate by the backbone network.