Abstract: It may be desirable for a remote advisor to request an autonomous vehicle to relocate to a target relocation pose using a freespace planner that has fewer constraints than the structured planner that is used for normal driving. The autonomous vehicle planning stack can include a remote vehicle interface, which can serve as the application programming interface for remote requestors to request relocation of the autonomous vehicle. The remote vehicle interface may implement a relocation state machine to carry out different parts of a relocation session. A scenario manager may subscribe to the state of the relocation state machine. Based on the state, the scenario manager can request the freespace planner to generate an output plan for the target relocation pose and can orchestrate handoff between different motion planners to relocate the autonomous vehicle.

Abstract: It may be desirable for a remote advisor to request an AV to relocate to a target relocation pose. In some cases, reaching a target relocation pose may involve a multi-point maneuver. Ability to perform relocation that involves a multi-point maneuver offers flexibility for the AV to maneuver around double-parked vehicles, to avoid obstacles, to aid a reroute when a route becomes unnavigable, etc. To support relocation that involves performing multi-point maneuvers, the planning stack may be modified to include a motion planner that can generate output plans that involve the AV going forward and in reverse. Moreover, the vehicle control stack may publish vehicle controls information as feedback information to the motion planner and the remote requestor, so that the motion planner and the remote requestor can be informed of the state of the AV.

Abstract: Autonomous vehicles utilize perception and understanding of road users and road objects to predict behaviors of the road users and road objects, and to plan a trajectory for the vehicle. Improved perception and understanding of the AV's surroundings can improve the AV's behavior around drivable objects, non-drivable road objects, construction zones, and temporary road closures. Improved perception and understanding of the AV's surroundings can also reduce the chances of the AV getting stuck and the need to be retrieved physically. To offer additional understanding capabilities, an additional understanding model is added to the perception and understanding pipeline to improve classification of road objects and extraction of attributes of the road objects. The implementation of the understanding model itself and placement of the model within the pipeline balance recall and precision performance metrics and computational complexity.

Abstract: One or more messages may be sent to a user who is to receive an item delivered by an AV. A message may include information of an attribute of the item that triggers the generation of the one or more messages. The attribute may indicate an extent of ease of the user carrying the item from a predetermined location, e.g., the drop-off location for the delivery. The attribute may be size, weight, temperature requirement, packaging, shape, etc. The item may be classified into a category requiring communication based on the attribute of the item and information of the user. A message may include a recommendation for using a tool or seeking help for carrying the item. A message may include an option to modify a setting (e.g., drop-off location) of the delivery. The classification of the item or generation of a message may be done by a machine learning model.

Abstract: Systems and methods for providing an individualized map highlighting undiscovered areas of a locale. In particular, systems and methods are provided for providing a map that visually differentiates areas a user has visited from areas a user has yet to explore. In some implementations, generalized maps are displayed with highlighted and lowlighted areas representing unexplored and explored areas of the map. A generalized map can show popular tourist destinations or other features of each area. In some examples, specific map details of a particular area are revealed when a user selects the particular map area. In some examples, unexplored areas are revealed when a user elects to visit the unexplored area.

Abstract: Autonomous vehicles utilize perception and understanding of road users to predict behaviors of the road users, and to plan a trajectory for the vehicle. Understanding subtypes and attributes of vehicles may help autonomous vehicles better predict behaviors of and react to vehicles in a variety of road situations. To offer additional understanding capabilities, an additional understanding model is added to the perception and understanding pipeline to improve classification of vehicles and extraction of attributes of the vehicles. The exemplary architectures of the understanding model balance recall and precision performance metrics and computational complexity.

Abstract: Disclosed are embodiments for facilitating a routing recommendation system based on user activities. In some aspects, an embodiment includes receiving input data comprising a list of activities requested by a user of an autonomous vehicle (AV) ride-hailing service and a point of destination requested by the user; mapping locations of interest corresponding to the list of activities; generating an ordered list of the locations of interest based on a user profile and historical data of the user; supplementing the ordered list of the locations of interest based on other historical data of other users of the AV ride-hailing service to generate a recommended list of places; applying a K-means clustering technique to the recommended list of places to identify one or more sets of recommendations and a centroid location for each set of recommendations; and providing a routing recommendation corresponding to a selected set of the sets of recommendations.

Abstract: Autonomous vehicles utilize perception and understanding of road users and road objects to predict behaviors of the road users and road objects, and to plan a trajectory for the vehicle. Understanding subtypes and attributes of vulnerable road users may help autonomous vehicles better predict behaviors of and react to vulnerable road users. To offer additional understanding capabilities, an additional understanding model is added to the perception and understanding pipeline to improve classification of vulnerable road users and extraction of attributes of the vulnerable road users. The exemplary architectures of the understanding model balance recall and precision performance metrics and computational complexity.

Abstract: Autonomous vehicles utilize perception and understanding of vehicles to predict behaviors of the vehicles, and to plan a trajectory. Understanding of attributes of vehicles may be improved through sensor fusion. Sensor fusion can be computationally expensive and may be difficult to implement in a real-time vehicle understanding system. To limit computational complexity while benefiting from machine learning across modalities, sensor fusion may be selectively implemented for a subset of task groups of a multi-task machine learning model. In some cases, part-based understanding may be implemented before fusion to limit the features being fused together to part features that are most salient for the task group. In addition, sensor data and features that may be fused together can be limited to sensor data and features within a desired field of view. A model that implements sensor fusion may be disabled for objects that are beyond a threshold distance.

Abstract: Tools can generate simulated driving tests that allow a same driving scenario to be tested at different geographical locations and with different arrangements of simulated road user(s) and/or simulated road objects. Numerous parameters used by these tools for creating the simulated driving tests can impact the usefulness of the simulated driving tests. If the parameters are not set appropriately, the simulated driving tests may not result in critical interactions that would enable accurate assessment of vehicle software. To alleviate some of these concerns, a parameter tuning system can find optimal value ranges of parameters based on criticality scores of generated simulated driving tests, so that better simulated driving tests can be obtained.

Abstract: Systems and methods for providing vehicle information to operators and service people without the use of a tool enabling communication directly between the vehicle and the operator or service person. In particular, systems and methods are provided for receiving a vehicle data request from an operator mobile device application, identifying a vehicle in close proximity to the operator mobile device, obtaining most recent vehicle data, and transmitting the requested vehicle data to the operator mobile device. Additionally, systems and methods are provided for the operator to request various capabilities with respect to the vehicles, such as vehicle access.

Abstract: Systems and methods for determining metrics for lane closure zone detection and evaluating the performance of lane closure zone detection. A lane closure zone can include any area where a lane is closed off. The changes to the typical travel paths for vehicles on the roadway can indicate that vehicles are to travel along a different path than that indicated on the maps the AVs use to operate. Thus, it is important for accurate driving operations of the AVs for the AVs to be able to interpret the changes to the typical travel paths indicated by the lane closure. Navigating through lane closure zones generally includes mapping (by the AV) objects in the lane closure zone, including the objects indicating the lane closure, and understanding related surrounding non-player characters (e.g., other vehicles).

Abstract: Machine learning models for controlling AV operations may be evaluated based on sliced data. A feature category critical for an environmental condition may be identified. The environmental condition is a condition in an environment where an AV performs an operation. A sub-dataset, which comprises sensor data capturing the environmental feature dataset, may be extracted from a dataset that comprises sensor data capturing the environment. The sub-dataset may be input into machine learning models that can classify the feature category. Performances of the machine learning models can be evaluated based on their outputs, which are generated based on the sub-dataset, and a ground-truth classification of the environmental feature. The output of each machine learning model comprises a classification of the environmental feature. A machine learning model may be selected based on the evaluated performances of the machine learning models and may be used to control driverless operations of AVs.

Abstract: Systems and methods for vehicles to detect and react to scene changes during stops for passenger pick-up, passenger drop-off, and deliveries. In particular, systems and methods are provided for a vehicle to detect a scene change while stopped and move a short distance from its stopped location. In general, after a vehicle has stopped to pick up or drop off a passenger or delivery, the vehicle remains stopped until it begins its drive to a next location.

Abstract: A volunteering platform facilitates voluntary activities associated with AV rides. The platform may receive a request for AV service from a first user and an offer to assist other users from a second user. The platform may determine whether an assistance to be provided by the second user can facilitate the service to be received by the first user. After determining that the assistance to be provided by the second user can facilitate the service to be received by the first user, the platform may pair the first user with the second user. The platform may select an AV from a fleet of AVs to serve the request from the first user with the assistance from the second user. The platform may plan an operation of the AV based on the request from the first user and the offer from the second user. The platform can dispatch the AV.

Abstract: Systems and methods for determining metrics for construction zone detection and evaluating the performance of construction zone detection. A construction zone can be manually labeled, and the human labels are converted to a format that can be input to a neural network. The neural network input represents polygons indicating the construction zone, as well as various construction zone objects. When an AV drives to the construction zone, the AV detects the construction zone objects and generates a predicted construction zone, including, for example, predicted edges. The predicted connectivity of the construction zone objects is evaluated to determine accuracy of the detected construction zone.

Abstract: A polyhedral sensor target includes multiple surfaces. A housing, such as a vehicle, may include a camera and a distance measurement sensor, such as a light detection and ranging (LIDAR) sensor. The housing may move between different positions during calibration, for instance by being rotated atop a turntable. The camera and distance measurement sensor may both capture data during calibration, from which visual and distance measurement representations of the polyhedral sensor target are identified. The camera and distance measurement sensor are calibrated based on their respective representations, for example by mapping vertices within the representations to the same location.

Abstract: Particular embodiments described herein provide for a system and method to help prevent condensation inside an enclosure that includes an electronic system. The system and method can determine a current relative humidity of an environment inside an enclosure that houses the electronic system, use the current relative humidity and the current maximum temperature to determine a dew point for the environment inside the enclosure, activate one or more heating elements inside the enclosure when the current minimum temperature is the same or lower than the dew point for the environment inside the enclosure, activate one or more embedded fans, and open one or more vents to help purge moisture out of the system during heating. When the current minimum temperature is higher than the dew point for the environment inside the enclosure, the system and method can allow the electronic system to power on.

Abstract: Disclosed are systems, apparatuses, methods, and computer-readable media to autonomous driving vehicles and, in particular, for tracking objects in an environment that an autonomous vehicle (AV) is navigating. A method includes receiving environment data from at least one sensor in an AV control system mounted to the AV, the environment data including online tracking data that identifies at least one object within the environment data that is recorded at drive time; annotating the at least one object from the environment data that are incorrectly identified by the AV control system; executing an offline tracking engine to generate offline tracking data that tracks the objects over time in the environment data; and identifying safety gaps between the online tracking data and the offline tracking data.

Abstract: The subject disclosure relates to techniques for filtering perception-related artifacts. The disclosed technology can include receiving, by a machine learning model, a first output generated by a perception system model onboard an autonomous vehicle, wherein the first output is based on sensor data received from sensors of the autonomous vehicle at a first time and includes an inaccurate perception of an environment around the autonomous vehicle, receiving, by the machine learning model, a second output generated by the perception system model onboard the autonomous vehicle, wherein the second output is based on sensor data received from sensors of the autonomous vehicle at a second time after the first time and includes an accurate perception of the environment around the autonomous vehicle, and altering, by the machine learning model, the inaccurate perception of the environment from the first output based on the accurate perception of the environment in the second output.