Abstract: Techniques are provided for foreground extraction from a point cloud (e.g., a LiDAR point cloud) using surface fitting. In an embodiment, one or more processors of a vehicle can receive point cloud data from one or more vehicle sensors. The one or more processors can identify points in the point cloud data as foreground points or ground points using a spatial Kalman filter to capture changes in the terrain. A route or trajectory in a driving area for the vehicle can be generated using the identified foreground points and ground points. A vehicle controller can control the vehicle while the vehicle is traveling on the route or trajectory in the driving area.

Abstract: Among other things, techniques are described for calibrating a sensor of a vehicle. One technique involves identifying a trigger for calibrating at least one sensor of a vehicle. In response to identifying the trigger, the technique then involves initiating a calibration path planning mode for the vehicle. In the calibration path planning mode, the technique involves: generating a plurality of calibration-aware paths that each include at least one calibration trajectory along at least one road; and selecting, from the plurality of paths, a first calibration-aware path for the vehicle, where the at least one sensor is calibrated while the vehicle travels along the selected path.

Abstract: Among other things, techniques are described for operating an autonomous vehicle station. One technique involves receiving information indicating arrival of a vehicle at a station designated for a primary service. The technique further involves measuring, using at least one sensor located in the station, a first parameter associated with the vehicle. Also, the technique involves performing, based on the information and the first parameter, a first action to provide the primary service to the vehicle. Additionally, the technique involves determining, while performing the first action, a secondary service to provide to the vehicle.

Abstract: A method for semantic annotation of sensor data may include obtaining sensor data representing an image of a geographic area. The boundary points defining a first polygon in the image of the geographic area may be determined based on the sensor data. An overlap between the first polygon and a second polygon in the image of the geographic area may be detected based at least on the boundary points defining the first polygon. At least one of the first polygon or the second polygon may be modified to remove the overlap between the first polygon and the second polygon. An annotation corresponding to the first polygon may be generated based on the modifying of at least one of the first polygon or the second polygon. The annotation may identify a physical feature within the geographic area. Related systems and computer program products are also provided.

Abstract: This disclosure relates in general to systems and methods for optically tracking objects proximate an autonomous vehicle. In particular, an object tracking system capable of refining position data for objects being tracked by determining a location of the objects surrounding the autonomous vehicle at least on part on previously determined locations of the objects. In certain instances, the predicted and detected locations used to arrive at a refined location for the objects can be weighted in different ways depending on conditions of the sensor data and quality of the historical data.

Abstract: Systems and methods for selecting a trajectory for an autonomous vehicle based on perceived risk are provided. A trajectory is selected that satisfies objective safety constraints and minimizes an overall cost based at least in part on cost components representing the perceived risk of the trajectory. The vehicle is navigated according to the selected (e.g., optimized) trajectory.

Abstract: Disclosed is a method including obtaining, using at least one processor, a user destination; obtaining, using the at least one processor, preference data indicative of a user route preference; determining, using the at least one processor, based on the user destination and the preference data, a set of routes towards a set of destinations, wherein at least one route of the set of routes is associated with one or more operational metrics; ranking, using the at least one processor, based on the preference data, the routes of the set of routes; and controlling, using the at least one processor, based on at least one of the ranked routes, navigation of an autonomous vehicle Systems and computer program products are also provided.

Abstract: The current disclosure provides techniques for using human driving behavior to assist in decision making of an autonomous vehicle as the autonomous vehicle encounters various scenarios on the road. For each scenario, a model may be generated based on human driving behavior that governs how an autonomous vehicle maneuvers in that scenario. As a result of using these models, reliability and safety of autonomous vehicle may be improved. In addition, because the model is programmed into the autonomous vehicle, the autonomous vehicle, in many instances, need not consume resources to implement complex calculations to determine driving behavior in real-time.

Abstract: Among other things, techniques are described for driver data guided spatial planning. A spatial structure is generated comprising a plurality of nodes connected by edges. At least some of the nodes and edges represent a path to navigate a vehicle from a first point to a second point. Edges of the spatial structure are labeled as useful based on a distance metric. The spatial structure is pruned by removing one or more edges from the spatial structure according to a respective label of the edges, wherein an extent of the removal is based on a predetermined graph size, a predetermined performance, or any combinations thereof to obtain a pruned graph. A path from the first point to the second point on the pruned graph is identified and the vehicle is navigated in accordance with the path from the first point to the second point on the pruned graph.

Abstract: Among other things, techniques are described for improving the trajectory estimates for an object in an environment. The technique includes receiving, by at least one processor, information indicating a presence of an object operating in an environment; determining, by the at least one processor, a trajectory of the object, the trajectory including at least a position, a speed, and a direction of travel of the object; determining, by the at least one processor, an expected route of the object, wherein the expected route is pre-planned and includes an expected future position of the object at a future time; comparing the trajectory of the object to the expected route of the object; and in accordance with the comparison that the trajectory of the object is consistent with the expected route of the object, updating the trajectory of the object based on the expected route of the object.

Abstract: A system can use semantic images, lidar images, and/or 3D bounding boxes to determine mobility parameters for objects in the semantic image. In some cases, the system can generate virtual points for an object in a semantic image and associate the virtual points with lidar points to form denser point clouds for the object. The denser point clouds can be used to estimate the mobility parameters for the object. In certain cases, the system can use semantic images, lidar images, and/or 3D bounding boxes to determine an object sequence for an object. The object sequence can indicate a location of the particular object at different times. The system can use the object sequence to estimate the mobility parameters for the object.

Abstract: The subject matter described in this specification is directed to systems and methods for dissipating heat from electronic components supporting autonomous vehicle systems. In particular, the specification describes how the electronic components can be positioned in a duct through which conditioned cabin air can be drawn to convectively dissipate heat from the electronic components.

Abstract: Among other things, we describe techniques for operation of a vehicle based on measured load characteristics and/or passenger comfort. One or more sensors of the vehicle can measure passenger data and/or load data of the vehicle. The passenger data and/or load data of the vehicle can be used by the vehicle to determine how to navigate within the surrounding environment.

Abstract: The present disclosure is directed to a computer system and techniques for automatically annotating objects in map data used for navigating an autonomous vehicle. Generally, the computer system is configured to obtain LiDAR data points for an environment around an autonomous vehicle, project the LiDAR data points onto image data, detect a target object in the image data, extract a subset of the LiDAR data points that corresponds to the detected target object, register the detected target object in map data if the extracted subset of the LiDAR data points satisfies registration criteria, and navigate the autonomous vehicle in the environment according to the map data.

Abstract: The subject matter described in this specification is directed to systems and methods for dissipating heat from sensors supporting autonomous vehicle systems. In particular, the specification describes how a housing, enclosing a sensor mounted to an exterior of a vehicle, can include heat dissipation components such as cooling fins across which conditioned cabin air and/or ambient air can be driven to convectively dissipate heat generated by the sensor.

Abstract: An automatic calibration and validation pipeline is disclosed to estimate and evaluate the accuracy of extrinsic parameters of a camera-to-LiDAR coordinate transformation. In an embodiment, an automated and unsupervised calibration procedure is employed where the computed rotational and translational parameters (“extrinsic parameters”) of the camera-to-LiDAR coordinate transformation are automatically estimated and validated, and upper bounds on the accuracy of the extrinsic parameters are set. The calibration procedure combines three-dimensional (3D) plane, vector and point correspondences to determine the extrinsic parameters, and the resulting coordinate transformation is validated by analyzing the projection of a filtered point cloud including a validation target in the image space. A single camera image and LiDAR scan (a “single shot”) are used to calibrate and validate the extrinsic parameters.

Abstract: Enclosed are embodiments for sampling driving scenarios for training machine learning models. In an embodiment, a method comprises: assigning, using at least one processor, a set of initial physical states to a set of objects in a map for a set of simulated driving scenarios, wherein the set of initial physical states are assigned according to one or more outputs of a random number generator; generating, using the at least one processor, the set of simulated driving scenarios in the map using the initial physical states of the objects in the set of objects; selecting, using the at least one processor, samples of the simulated driving scenarios; training, using the at least one processor, a machine learning model using the selected samples; and operating, using a control circuit, a vehicle in an environment using the trained machine learning model.

Abstract: In an example method, a computer system receives first data representing a plurality of scenarios for estimating a performance of a vehicle in conducting autonomous operations. Further, the computer system determines, for each of the scenarios: (i) a first metric indicating an observed performance of the vehicle in that scenario, where the first metric is determined based on at least one rule, and (ii) a second metric indicating a degree of information gain associated with that scenario. The computer system selects a subset of the scenarios based on the first metrics and the second metrics, and outputs second data indicative of the subset of the scenarios.

Abstract: Among other things, a system provides speed behavior planning for vehicles with autonomous driving capabilities.

Abstract: Among other things, techniques are described for notifying and explaining the action performed by an autonomous vehicle, including but not limited to: receiving a planned path of a vehicle, a state of the vehicle and environment data of an environment in which the vehicle is operating, receiving a deviation signal, determining whether the deviation signal was reported by a first system or a second system of the vehicle, in response selecting a first set of simulators or a second set of simulators for simulating the vehicle in the environment, simulating the vehicle in the environment using the selected first or second set of simulators, based on results of the simulating, generating a message and presenting the message to at least one occupant of the vehicle.