A method for training a machine learning algorithm for predicting an intent parameter of an object in proximity to a self-driving vehicle on a terrain are provided. The method includes generating a training dataset having assessor-less labels, based on data collected by a training vehicle. The data collected by the training vehicle include data on the state of the training vehicle, the state of a training object, and a training terrain at a target moment in time and at a time after the target moment in time. The training data is based, at least in part, on the data for the target moment in time, and the assessor-less label is based, at least in part, on the data for a time after the target moment in time. A method for operating a self-driving vehicle and a self-driving vehicle are also disclosed.

A transportation vehicle with at least one first sensor for capturing environment data, at least one second sensor for capturing transportation vehicle data, a communication module for establishing a data connection with another transportation vehicle, a driving system for automated driving of the transportation vehicle, at least one output element for a visible/audible warning signal, and a control unit. The control unit determines a predicted trajectory of the transportation vehicle, determines a predicted path of the transportation vehicle and receives a predicted trajectory and vehicle geometry data of the other transportation vehicle via the data connection, determines a predicted path of the other transportation vehicle, determines a possible collision of the transportation vehicle with the other transportation vehicle, and in response to a possible collision, outputs a warning signal by the at least one output element and/or carries out an automated driving maneuver by the driving system.

An autonomous vehicle uses machine learning based models to predict hidden context attributes associated with traffic entities. The system uses the hidden context to predict behavior of people near a vehicle in a way that more closely resembles how human drivers would judge the behavior. The system determines an activation threshold value for a braking system of the autonomous vehicle based on the hidden context. The system modifies a world model based on the hidden context predicted by the machine learning based model. The autonomous vehicle is safely navigated, such that the vehicle stays at least a threshold distance away from traffic entities.

Systems and methods are described that probabilistically predict dynamic object behavior. In particular, in contrast to existing systems which attempt to predict object trajectories directly (e.g., directly predict a specific sequence of well-defined states), a probabilistic approach is instead leveraged that predicts discrete probability distributions over object state at each of a plurality of time steps. In one example, systems and methods predict future states of dynamic objects (e.g., pedestrians) such that an autonomous vehicle can plan safer actions/movement.

A vehicle behavior generation device sets multiple possible behaviors of an own vehicle when the own vehicle travels along a planned route; sets multiple possible behaviors of a different vehicle existing around the own vehicle corresponding to each of the set multiple possible behaviors of the own vehicle; outputs information indicating a contact possibility between the own vehicle and the different vehicle for each of combinations of the set multiple possible behaviors of the own vehicle and the set multiple possible behaviors of the different vehicle; and selects one of the multiple possible behaviors of the own vehicle based on the outputted information.

An autonomous vehicle (AV) includes features that allows the AV to comply with applicable regulations and statues for performing safe driving operation. An example method for operating the AV includes determining a trajectory related information of a vehicle operating on a roadway on which the AV is operating; receiving sensor data of a first area that includes the vehicle; determining an additional trajectory related information for the AV by comparing the trajectory related information of the vehicle to a current trajectory related information of the AV, wherein the additional trajectory related information is based on a category to which the vehicle belongs, and wherein the additional trajectory related information allows the AV to maintain at least a distance between the AV and the vehicle; and causing the AV to operate in accordance with the additional trajectory related information.

A trajectory for an autonomous machine may be evaluated for safety based at least on determining whether the autonomous machine would be capable of occupying points of the trajectory in space-time while still being able to avoid a potential future collision with one or more objects in the environment through use of one or more safety procedures. To do so, a point of the trajectory may be evaluated for conflict based at least on a comparison between points in space-time that correspond to the autonomous machine executing the safety procedure(s) from the point and arrival times of the one or more objects to corresponding position(s) in the environment. A trajectory may be sampled and evaluated for conflicts at various points throughout the trajectory. Based on results of one or more evaluations, the trajectory may be scored, eliminated from consideration, or otherwise considered for control of the autonomous machine.

The methods and systems are directed to computational approaches for training and using machine learning algorithms to predict the conditional marginal distributions of the position of agents at flexible evaluation horizons and can enables more efficient path planning. These methods model agent movement by training a deep neural network to predict the position of an agent through time. A neural ordinary differential equation (neural ODE) that represents this neural network can be used to determine the log-likelihood of the agent's position as it moves in time.

A system for maneuvering a vehicle has a detection system, a prediction system, and a vehicle control system. The detection system is configured to detect a nearby vehicle adjacent to the vehicle. The prediction system is configured to calculate a predicted trajectory of the nearby vehicle upon receiving a detection result from the detection system. The vehicle control system is configured to maneuver the vehicle based on the predicted trajectory upon receiving a control signal from the prediction system. The vehicle control system maneuvers the vehicle to keep a specified distance away from the nearby vehicle. A method for maneuvering a vehicle includes detecting a nearby vehicle adjacent to the vehicle, calculating a predicted trajectory of the nearby vehicle, and maneuvering the vehicle based on the predicted trajectory to keep a specified distance away from the nearby vehicle.

Categorizing driving behaviors of other road users includes maintaining a first history of first lateral-offset values of a road user with respect to a center line of a lane of a road; determining a first pattern based on the first history of the first lateral-offset values; determining a driving behavior of the road user based on the first pattern; and autonomously performing, by a host vehicle, a driving maneuver based on the driving behavior of the road user. The first history can be maintained for a predetermined period of time. An apparatus includes a processor that is configured to track a trajectory history of a road user; determine, based on the trajectory history, a driving behavior of the road user; and transmit a notification of the driving behavior.