Abstract: A light detection and ranging (LIDAR) sensor assembly can comprise an optics assembly that includes a LIDAR sensor and a set of dovetail joint inserts. The LIDAR sensor assembly can further include a frame comprising a set of dovetail joint septums coupled to the set of dovetail joint inserts of the optics assembly.

Abstract: Disclosed are methods, systems, and non-transitory computer readable media that control an autonomous vehicle via at least two sensors. One aspect includes capturing an image of a scene ahead of the vehicle with a first sensor, identifying an object in the scene at a confidence level based on the image, determining the confidence level of the identifying is below a threshold, in response to the confidence level being below the threshold, directing a second sensor having a field of view smaller than the first sensor to generate a second image including a location of the identified object, further identifying the object in the scene based on the second image, controlling the vehicle based on the further identification of the object.

Abstract: An on-board computing system for a vehicle is configured to generate and selectively present a set of autonomous-switching directions within a navigation user interface for the operator of the vehicle. The autonomous-switching directions can inform the operator regarding changes to the vehicle's mode of autonomous operation. The on-board computing system can generate the set of autonomy-switching directions based on the vehicle's route and other information associated with the route, such as autonomous operation permissions (AOPs) for route segments that comprise the route. The on-board computing device can selectively present the autonomy-switching directions based on locations associated with anticipated changes in autonomous operations determined for the route of the vehicle, the vehicle's location, and the vehicle's speed. In addition, the on-board computing device is further configured to present audio alerts associated with the autonomy-switching directions to the operator of the vehicle.

Abstract: An autonomous vehicle (AV) includes a vehicle computing system configured to receive map data of a geographic location, obtain position estimate data of the autonomous vehicle and determine a route of the autonomous vehicle including a plurality of roadways in the plurality of submaps. The autonomous vehicle determines a route including a plurality of roadways, determines a first roadway in the plurality of roadways closest to the position estimate and a second roadway outside the plurality of roadways closest to the position estimate of the autonomous vehicle, and determines a pose relative to the first roadway or the second roadway based on a distance between the position estimate of the autonomous vehicle and a roadway associated with a prior pose of the autonomous vehicle to control travel of the autonomous vehicle based on the vehicle pose.

Abstract: Systems, device, and methods for trajectory association and tracking are provided. A method can include obtaining input data indicative of a respective trajectory for each of one or more first objects for a first time step and input data indicative of a respective trajectory for each of one or more second objects for a second time step subsequent to the first time step. The method can include generating, using a machine-learned model, a temporally-consistent trajectory for at least one of the one or more first objects or the one or more second objects based at least in part on the input data and determining a third predicted trajectory for the at least one of the one or more first objects or the one or more second objects for at least the second time step based at least in part on the temporally-consistent trajectory.

Abstract: Example aspects of the present disclosure describe a scene generator for simulating scenes in an environment. For example, snapshots of simulated traffic scenes can be generated by sampling a joint probability distribution trained on real-world traffic scenes. In some implementations, samples of the joint probability distribution can be obtained by sampling a plurality of factorized probability distributions for a plurality of objects for sequential insertion into the scene.

Abstract: The present disclosure provides systems and methods to obtain feedback descriptive of autonomous vehicle failures. In particular, the systems and methods of the present disclosure can detect that a vehicle failure event occurred at an autonomous vehicle and, in response, provide an interactive user interface that enables a human located within the autonomous vehicle to enter feedback that describes the vehicle failure event. Thus, the systems and methods of the present disclosure can actively prompt and/or enable entry of feedback in response to a particular instance of a vehicle failure event, thereby enabling improved and streamlined collection of information about autonomous vehicle failures.

Abstract: Systems, methods, tangible, non-transitory computer-readable media, and devices for configuration of a vehicle compartment are provided. For example, a method can include receiving, by a computing system, occupancy data based in part on one or more states of one or more objects. Based in part on the occupancy data and compartment data, a compartment configuration can be determined for one or more compartments of an autonomous vehicle. The compartment data can be based in part on a state of the one or more compartments. The compartment configuration can specify one or more spatial relations of one or more compartment components associated with the one or more compartments. One or more configuration signals can be generated based in part on the compartment configuration to control the one or more compartments of the autonomous vehicle.

Abstract: Systems and methods for generating object segmentations across videos are provided. An example system can enable an annotator to identify objects within a first image frame of a video sequence by clicking anywhere within the object. The system processes the first image frame and a second, subsequent, image frame to assign each pixel of the second image frame to one of the objects identified in the first image frame or the background. The system refines the resulting object masks for the second image frame using a recurrent attention module based on contextual features extracted from the second image frame. The system receives additional user input for the second image frame and uses the input, in combination with the object masks for the second image frame, to determine object masks for a third, subsequent, image frame in the video sequence. The process is repeated for each image in the video sequence.

Abstract: Systems and methods for basis path generation are provided. In particular, a computing system can obtain a target nominal path. The computing system can determine a current pose for an autonomous vehicle. The computing system can determine, based at least in part on the current pose of the autonomous vehicle and the target nominal path, a lane change region. The computing system can determine one or more merge points on the target nominal path. The computing system can, for each respective merge point in the one or more merge points, generate a candidate basis path from the current pose of the autonomous vehicle to the respective merge point. The computing system can generate a suitability classification for each candidate basis path. The computing system can select one or more candidate basis paths based on the suitability classification for each respective candidate basis path in the plurality of candidate basis paths.

Abstract: Systems and methods are provided for machine-learned models including convolutional neural networks that generate predictions using continuous convolution techniques. For example, the systems and methods of the present disclosure can be included in or otherwise leveraged by an autonomous vehicle. In one example, a computing system can perform, with a machine-learned convolutional neural network, one or more convolutions over input data using a continuous filter relative to a support domain associated with the input data, and receive a prediction from the machine-learned convolutional neural network. A machine-learned convolutional neural network in some examples includes at least one continuous convolution layer configured to perform convolutions over input data with a parametric continuous kernel.

Abstract: Systems and methods for automatically servicing autonomous vehicles are provided. In one example embodiment, a computer implemented method includes obtaining data associated with one or more reference mechanisms located on an autonomous vehicle. The method includes identifying information associated with the autonomous vehicle based at least in part on the data associated with the one or more reference mechanisms located on the autonomous vehicle. The information associated with the autonomous vehicle includes an orientation of the autonomous vehicle. The method includes determining a vehicle maintenance plan for the autonomous vehicle based at least in part on the information associated with the autonomous vehicle. The method includes providing one or more control signals to implement the vehicle maintenance plan for the autonomous vehicle based at least in part on the orientation of the autonomous vehicle.

Abstract: In one example embodiment, a computer-implemented method for coordinating the movement of assets at transfer hubs includes determining one or more assets to move from a first location associated with a transfer hub. The method also includes assigning a jockey to move the one or more assets to a second location associated with the transfer hub based at least in part on an availability of the jockey. The method further includes directing the jockey to move the one or more assets from the first location to the second location.

Abstract: Example aspects of the present disclosure are directed to improved systems and methods for testing autonomous vehicle operation through the use of mobile test devices that are controlled at least in part in response to predictive motion planning associated with autonomous vehicles. More particularly, the motion of a mobile test device can be controlled based on the motion plan of an autonomous vehicle to cause desired interactions between the mobile test device and the autonomous vehicle. Motion planning data associated with the autonomous vehicle can be obtained by an autonomous vehicle (AV) test system prior to the autonomous vehicle implementing or completely implementing a motion plan. In this manner, the AV test system can proactively control the mobile test device based on predictive motion planning data to facilitate interactions between the mobile test device and the autonomous vehicle that may not otherwise be achievable.

Abstract: Systems, methods, tangible non-transitory computer-readable media, and devices associated with object perception and prediction of object motion are provided. For example, a plurality of temporal instance representations can be generated. Each temporal instance representation can be associated with differences in the appearance and motion of objects over past time intervals. Past paths and candidate paths of a set of objects can be determined based on the temporal instance representations and current detections of objects. Predicted paths of the set of objects using a machine-learned model trained that uses the past paths and candidate paths to determine the predicted paths. Past path data that includes information associated with the predicted paths can be generated for each object of the set of objects respectively.

Abstract: The present disclosure provides systems and methods for generating photorealistic image simulation data with geometry-aware composition for testing autonomous vehicles. In particular, aspects of the present disclosure can involve the intake of data on an environment and output of augmented data on the environment with the photorealistic addition of an object. As one example, data on the driving experiences of a self-driving vehicle can be augmented to add another vehicle into the collected environment data. The augmented data may then be used to test safety features of software for a self-driving vehicle.

Abstract: The present disclosure provides systems and methods that combine physics-based systems with machine learning to generate synthetic LiDAR data that accurately mimics a real-world LiDAR sensor system. In particular, aspects of the present disclosure combine physics-based rendering with machine-learned models such as deep neural networks to simulate both the geometry and intensity of the LiDAR sensor. As one example, a physics-based ray casting approach can be used on a three-dimensional map of an environment to generate an initial three-dimensional point cloud that mimics LiDAR data. According to an aspect of the present disclosure, a machine-learned model can predict one or more dropout probabilities for one or more of the points in the initial three-dimensional point cloud, thereby generating an adjusted three-dimensional point cloud which more realistically simulates real-world LiDAR data.

Abstract: Aspects of the present disclosure involve systems, methods, and devices for fault detection in a Lidar system. A fault detection system obtains incoming Lidar data output by a Lidar system during operation of an AV system. The incoming Lidar data includes one or more data points corresponding to a fault detection target on an exterior of a vehicle of the AV system. The fault detection system accesses historical Lidar data that is based on data previously output by the Lidar system. The historical Lidar data corresponds to the fault detection target. The fault detection system performs a comparison of the incoming Lidar data with the historical Lidar data to identify any differences between the two sets of data. The fault detection system detects a fault condition occurring at the Lidar system based on the comparison.

Abstract: The present disclosure provides systems and methods that enable human supervision of a highly capable automated driving system. In particular, the systems and methods of the present disclosure enable a human (e.g., a passenger, driver/operator, or remote supervisor of an autonomous vehicle) to easily and quickly transition control of the autonomous vehicle from a primary motion plan that controls the vehicle towards a primary destination to a secondary motion plan that controls the vehicle to a safe state. As such, the systems and methods of the present disclosure enable advanced human supervision of autonomous vehicle behavior in which a human can cause an autonomous vehicle to operate in a risk-reduced manner or otherwise maneuver to a safe state, without requiring the human to actually assume manual control of the vehicle.

Abstract: A sensor data processing system for an autonomous vehicle receives inertial measurement unit (IMU) data from one or more IMUs of the autonomous vehicle. Based at least in part on the IMU data, the system identifies an IMU data offset from a deficient IMU of the one or more IMUs, and generates an offset compensation transform to compensate for the IMU data offset from the deficient IMU. The system dynamically executes the offset compensation transform on the IMU data from the deficient IMU to dynamically compensate for the IMU data offset.