A control method for a vehicle (101) in which the vehicle (101) is controlled manually while sensor data, from at least one sensor of the vehicle (101), is collected and stored. Then the vehicle (101) is controlled autonomously while sensor data, from the at least one sensor is detected and matched to the stored sensor data.

A fleet management system dispatches autonomous electric vehicles (AEVs) as on-demand power sources. The fleet management system receives a request for a power source including a location and data describing the amount of power requested. The fleet management system selects an AEV of the fleet to service the request based on the relative locations of the AEVs to the requested location, and based on the amount of power requested. The fleet management system instructs the selected AEV to drive to the location and supply power. The fleet management system instructs the selected AEV to disconnect and return to the charging station, and may instruct another AEV to continue fulfilling the request if additional power is needed.

Techniques are discussed for determining prediction probabilities of an object based on a top-down representation of an environment. Data representing objects in an environment can be captured. Aspects of the environment can be represented as map data. A multi-channel image representing a top-down view of object(s) in the environment can be generated based on the data representing the objects and map data. The multi-channel image can be used to train a machine learned model by minimizing an error between predictions from the machine learned model and a captured trajectory associated with the object. Once trained, the machine learned model can be used to generate prediction probabilities of objects in an environment, and the vehicle can be controlled based on such prediction probabilities.

Aspects of the disclosure relate to a system that includes a memory storing a queue for arranging tasks, a plurality of self-driving systems for controlling an autonomous vehicle, and one or more processors. The one or more processors may receive a non-passenger task request with a priority level of the non-passenger task request. When the non-passenger task request is accepted, the one or more processors may insert the task in the queue based on the priority level of the task request. Then, the one or more processors may provide instructions to one or more self-driving systems according to the non-passenger task request. Having received updates of the status of the autonomous vehicle, the one or more processors may determine that the task is completed based on the updates. After determining that the task is completed, the one or more processors may remove the task from the queue.

To improve prediction accuracy in model predictive control. The ship control device includes processing circuitry. The processing circuitry estimates an initial search value of a throttle opening by an estimation method based on uncertainty with a distance between a position of the ship and a target position of dynamic positioning, and a true wind velocity as preconditions. The processing circuitry searches an action for moving the ship to the target position by a model predictive control in a search range having the initial search value as an origin, and determines a command throttle opening based on the search result.

Data processing systems disclosed herein may promote satisfaction of a rider of a vehicle and include a machine learning model and a vehicle control system. The machine learning model determines a measure of an emotional state of the rider based on data received from a sensor associated with the rider, where the data is indicative of a physiological condition of the rider. The vehicle control system: determines a target value of an operating parameter of the vehicle based on a correlation between the emotional state of the rider and the target value of the operating parameter; and adjusts the operating parameter of the vehicle based on the target value of the operating parameter.

An unmanned vehicle control method includes acquiring a delivery request for an item, the delivery request comprising delivery information of the item, and determining, according to the delivery information, predicted travelling data associated with delivering the item and at least one of network coverage or network connection quality associated with the predicted travelling data. The method further includes allocating network resources according to the at least one of the network coverage or the network connection quality of the predicted travelling data, and generating a remote driving control instruction according to the predicted travelling data. The method further includes transmitting the remote driving control instruction to an unmanned vehicle using the allocated network resources, so as to cause the unmanned vehicle to drive based on the remote driving control instruction, the unmanned vehicle being configured to transport the item.

A method is provided for causing one or more robots to execute a mission. The method includes determining a behavior tree in which the mission is modeled, and causing the one or more robots to execute the mission using the behavior tree and a leaf node library. The behavior tree is expressed as a directed tree of nodes including a switch node, a trigger node representing a selected task, and action nodes representing others of the tasks. The switch node is connected to the trigger node and the action nodes in a parent-child relationship in which the trigger node and the action nodes are children of the switch node. The trigger node is a first of the children that, when ticked by the switch node, returns an identifier of one of the action nodes to trigger the switch node to next tick the one of the action nodes.

Techniques for latent belief space planning include: during execution of an autonomous agent configured to control operation of a physical mechanism, obtaining a current observation of a physical environment; based at least on the current observation of the physical environment, generating a trajectory tree that represents possible trajectories in a belief space, wherein nodes of the trajectory tree represent values of a continuous observation, a continuous state, and a continuous control, each node being associated with one of multiple timesteps along the plurality of possible trajectories, and wherein branches from inner nodes to child nodes correspond to possible outcomes and observations of a multi-modal latent state; determining a current value of the continuous control associated with a current node; and applying the current value of the continuous control to operation of the physical mechanism.

A system for transportation includes a vehicle interface for gathering physiological sensed data of a rider in the vehicle. The system includes an artificial intelligence-based circuit that is trained on a set of outcomes related to rider in-vehicle experience and that induces, responsive to the sensed rider physiological data, variation in one or more of the user experience parameters to achieve at least one desired outcome in the set of outcomes. The inducing variation includes control of timing and extent of the variation.