Abstract: Techniques for using a detector coupled to a controller area network (CAN) bus of a vehicle system to identify and generate alerts in response to wake-up attacks on the vehicle system. Techniques include an electronic control unit (ECU) sending a wake-up message across the CAN bus that is detected by the detector. The detector includes memory and a processor to identify the timestamp of the wake-up message when the vehicle ignition is off. The detector determines a total operational time for the ECU over an observation time period and generates a notification of a wake-up anomaly when the total operational time over the observation time period exceeds a predetermined threshold.

Abstract: A method and apparatus for controlling a vehicle are provided. A lane in which a target vehicle is driving, and an object in a vicinity of the lane are detected from an image of surroundings of the target vehicle, a degree of danger of the object is evaluated, driving information of the target vehicle is determined based on the degree of danger, and the target vehicle is controlled based on the driving information.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a neural network having a plurality of sub neural networks to assign respective confidence scores to one or more candidate future trajectories for an agent. Each confidence score indicates a predicted likelihood that the agent will move along the corresponding candidate future trajectory in the future. In one aspect, a method includes using the first sub neural network to generate a training intermediate representation; using the second sub neural network to generate respective training confidence scores; using a trajectory generation neural network to generate a training trajectory generation output; computing a first loss and a second loss; and determining an update to the current values of the parameters of the first and second sub neural networks.

Abstract: A charging control system includes a lawn mower that has a battery and performs a lawn mowing work while traveling autonomously, and a charging station for charging the battery. The lawn mower includes a period calculator for calculating a shutoff period of supply power supplied from the charging station, and a first communication unit. The charging station includes a second communication unit communicating with the first communication unit, an information acquisition unit for acquiring shutoff period information indicating the shutoff period from the first communication unit via the second communication unit, a switch for shutting off the supply power, and a shutoff controller for controlling the operation of the switch. The shutoff controller releases the shutoff of the power supply to the lawn mower based on the shutoff period information.

Abstract: The technology disclosed provides a so-called “pointer sentinel mixture architecture” for neural network sequence models that has the ability to either reproduce a token from a recent context or produce a token from a predefined vocabulary. In one implementation, a pointer sentinel-LSTM architecture achieves state of the art language modeling performance of 70.9 perplexity on the Penn Treebank dataset, while using far fewer parameters than a standard softmax LSTM.

Abstract: A system and method to distribute pesticides, fertilizers, water, and other materials on a farm with accuracy and precision is disclosed in order to combat the problems imposed on the environment due to over-fertilization and over use of pesticides. This system and method is a social networking control system in which multiple farms have independent grids of sensors capable of detecting the presence of pesticides, fertilizers, water, and other materials in the air, in the top-soil, and in the groundwater. These grids of sensors detect the location and concentration of these materials and reports them back to a social control system for analysis. The control system regulates the deposition of further chemicals through computer control of the chemical dispersal systems.

Abstract: A method, apparatus, and system for planning a deceleration trajectory based on a natural deceleration profile for an autonomous driving vehicle (ADV) is disclosed. In one embodiment, in response to a request for driving in a natural deceleration mode, a current speed of the ADV is determined. A speed guideline is generated based on a predetermined natural deceleration profile associated with the ADV in view of the current speed of the ADV. A speed planning operation is performed by optimizing a total cost function based on the speed guideline to determine speeds of a plurality of trajectory points along a trajectory planned to drive the ADV. A number of control commands are generated based on the speed planning operation to control the ADV with the planned speeds along the trajectory, such that the ADV naturally slows down according to the predetermined natural deceleration profile.

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: The present disclosure provides a method comprising identifying at least one of a characteristic and an identity of an item for delivery from an origin to a destination; identifying a plurality of possible routes between the origin and the destination using mapping information, the mapping information including for each of the plurality of possible routes, a characterization of each of a plurality of route segments comprising the possible route; evaluating the plurality of possible routes in view of the identified at least one of the item characteristic and the item identity to select one of the plurality of possible routes; and providing the selected one of the plurality of possible routes to a vehicle, wherein the vehicle delivers the item from the origin to the destination via the identified route.

Abstract: A method and apparatus for vehicle damage assessment, an electronic device, and a computer-readable storage medium are provided. The method may include: extracting, from an input image, a first feature characterizing a part of a vehicle and a second feature characterizing a damage type of the vehicle; integrating the first feature and the second feature to generate a third feature characterizing a corresponding relation between the part and the damage type; converting the third feature into a characteristic vector; and determining a damage recognition result based on the characteristic vector. According to the technical solution of the disclosure, users can rapidly and accurately learn about the damage condition of the vehicle by providing pictures or videos of the damaged vehicle, thus providing an objective basis for subsequent damage assessment, claim settlement, and repair.

Abstract: Systems and methods provide for enabling an autonomous vehicle to provide road assistance to a vehicle. The autonomous vehicle can analyze sensor data about the vehicle captured by one or more of its sensors as it navigates a route. Based on the analysis of the sensor data, the autonomous vehicle can determine that the vehicle needs maintenance. The autonomous vehicle can proactively send a request, based on the determination, to initiate a towing mechanism to tow the vehicle. Based on receiving an acceptance of the request from the vehicle, the autonomous vehicle can activate the towing mechanism.

Abstract: Provided is a method for detecting a driver's abnormalities based on a CAN (Controller Area Network) bus network communicating with an ECU (Electronic Control Unit) of a vehicle. The method may include: acquiring a CAN bus signal related to an operation of the vehicle from the CAN bus network; extracting a detection vector from the CAN bus signal using an auto encoder; and detecting a driver's abnormality based on the detection vector.

Abstract: Among other things, we describe techniques for an interface board for harmonizing performance of different autonomous vehicles in a fleet. In an embodiment, the interface system includes a printed circuit board installed in a host vehicle that includes an interface controller and interface circuitry coupled to the interface controller. The interface circuitry includes a plurality of relays configured by the interface controller to route/pass sensor signals received from actuators of the host vehicle to the interface controller, and route/pass control signals from the interface controller to actuator controllers of the host vehicle that control the actuators in accordance with an operating mode.

Abstract: Systems and methods for controlling autonomous vehicles are provided. Assisted autonomy tasks facilitated by operators for a plurality of autonomous vehicles can be tracked in order to generate operator attributes for each of a plurality of operators. The attributes for an operator can be based on tracking one or more respective assisted autonomy tasks facilitated by the operator. The operator attributes can be used to facilitate enhanced remote operations for autonomous vehicles. For example, request parameters can be obtained in response to a request for remote assistance associated with an autonomous vehicle. An operator can be selected to assist with autonomy tasks for the autonomous vehicle based at least in part on the operator attributes for the operator and the request parameters associated with the request. Remote assistance for the first autonomous vehicle can be initiated, facilitated by the first operator in response to the request for remote assistance.

Abstract: A method and system for remote vehicle diagnostics are provided. The method is applicable to a remote vehicle diagnostic system including a vehicle diagnostic terminal and a vehicle connector. A remote communication connection between the vehicle diagnostic terminal and the vehicle connector is established via a preset programming interface. The vehicle diagnostic terminal receives diagnostic instruction data of a diagnostic application. The vehicle diagnostic terminal transmits the diagnostic instruction data to the vehicle connector via the preset programming interface by way of the remote communication. The vehicle connector converts the diagnostic instruction data into first data in a first data format, and performs a diagnostic operation corresponding to the first data, where the first data format is readable by a vehicle.

Abstract: Techniques are discussed for controlling a vehicle, such as an autonomous vehicle, based on occluded areas in an environment. An occluded area can represent areas where sensors of the vehicle are unable to sense portions of the environment due to obstruction by another object or sensor limitation. An occluded region for an object is determined by the vehicle as part of an occlusion grid, from the perspective of the vehicle. The vehicle may receive another occlusion grid from another source, such as another vehicle or a remote computing device that stores and distributes occlusion grids. The other occlusion grid may be from a different perspective than the occlusion grid generated by the vehicle, and may include occupancy data for the region that is otherwise occluded from the perspective of the vehicle. The vehicle can be controlled to traverse the environment based on the occupancy data received from the other source.

Abstract: A method includes receiving, at a mobile hub device, communications including location-specific risk data and a task assignment. The method also includes generating an output indicating dispatch coordinates. The dispatch coordinates identifying a dispatch location from which to dispatch, from the mobile hub device, one or more unmanned vehicles to perform a task of the task assignment.

Abstract: Graphs are powerful structures made of nodes and edges. Information can be encoded in the nodes and edges themselves, as well as the connections between them. Graphs can be used to create manifolds which in turn can be used to efficiently train more robust AI systems. Systems and methods for graph-based AI training in accordance with embodiments of the invention are illustrated. In one embodiment, a graph interface system including a processor, and a memory configured to store a graph interface application, where the graph interface application directs the processor to obtain a set of training data, where the set of training data describes a plurality of scenarios, encode the set of training data into a first knowledge graph, generate a manifold based on the first knowledge graph, and train an AI model by traversing the manifold.

Abstract: A method, a computer program with instructions, and an apparatus for configuring a control system for an at least partially autonomous transportation vehicle. Data relating to the driving situation are captured, selection criteria are then determined from the available data, two or more AI modules are selected from a library of AI modules during driving based on selection criteria, and combined execution of the selected two or more AI modules by the control system is initiated.

Abstract: This application relates to an abstract description generating method, an abstract description generation model training method, a computer device, and a storage medium. The abstract description generating method includes: inputting a labeled training sample into an abstract description generation model; performing first-phase training on an encoding network and a decoding network of the abstract description generation model based on supervision of a first loss function; obtaining a backward-derived hidden state of a previous moment through backward derivation according to a hidden state of each moment outputted by the decoding network; obtaining a value of a second loss function according to the backward-derived hidden state of the previous moment and an actual hidden state of the previous moment outputted by the decoding network; and obtaining final model parameters of the abstract description generation model determined based on supervision of the second loss function to reach a preset threshold value.