Gm Global Technolgoy Operations, Inc.

Roi Reshef

Vlad Goldner

Avshalom Suissa

Zahy Bnaya

Yevgeni Shaphirov

Examples described herein provide a computer-implemented method for processing environmental data for at least one other vehicle relative to a target vehicle. The method includes processing, by a processing device using a self-attention mechanism, first data associated the at least one other vehicle to generate a first output associated with the at least one other vehicle. The method further includes processing, by the processing device using stacked residual blocks, second data associated the target vehicle to generate a second output associated with the target vehicle. The method further includes concatenating, by the processing device, the first output and the second output to generate a concatenated vector. The method further includes making, by the processing device, a driving decision for the target vehicle based at least in part on the concatenated vector. The method further includes controlling the target vehicle based at least in part on the decision.

Processing environmental data for vehicles

Methods and systems are provided for controlling an autonomous vehicle. In one embodiment, a method includes: identifying, by a processor, at least one constraint on a longitudinal dimension of an upcoming road; defining, by the processor, constraint activation logic based on a type of the at least one constraint; performing, by the processor, the constraint activation logic to determine a state of the constraint to be at least one of active and inactive; when the state of the constraint is active, validating, by the processor, a motion plan of the autonomous vehicle based on the constraint; and selectively controlling the autonomous vehicle based on the validating of the motion plan.

Autonomous look ahead methods and systems

A vehicle, and a system a method of navigating a vehicle. The system includes a trajectory planning module and a trajectory tracking module. The trajectory planning module operates at a processor of the vehicle to generate a trajectory for the vehicle. The trajectory tracking module operates at the processor to track the trajectory to navigate the vehicle. The trajectory planning module and the trajectory tracking module run asynchronously from each other.

Method of decoupling trajectory planning and tracking

Systems and methods to perform behavioral planning in an autonomous vehicle from a reference state involve generating a set of actions of a fixed size and fixed order according to a predefined methodology. Each action is a semantic instruction for a next motion of the vehicle. A set of trajectories is generated from the set of actions as an instruction indicating a path and a velocity profile to generate steering angles and accelerations for implementation by the vehicle. A trajectory filter is applied to filter the set of trajectories such that unfiltered trajectories are candidate trajectories. Applying the trajectory filter includes assessing the path and velocity profile indicated by each of the set of trajectories. A selected trajectory is used to control the vehicle or the action that corresponds to the selected trajectory is used in trajectory planning to generate a final trajectory that is used to control the vehicle.

Behavioral planning in autonomus vehicle

Examples described herein provide a computer-implemented method that includes receiving, by a processing device, a current state of a vehicle. The method further includes predicting, by the processing device using an output of an artificial intelligence model, a future state of the vehicle based at least in part on the current state of the vehicle. The method further includes calculating, by the processing device using a tunable reward function, a reward associated with the future state of the vehicle, the tunable reward function comprising multiple tunable coefficients. The method further includes training, by the processing device, the artificial intelligence model based at least in part on the reward.

Reward function for vehicles

A system and method operate an autonomous vehicle. A sensor senses a road and an object. A processor determines, in a Cartesian reference frame, a representation of the road and a source point representative of the object, samples a first waypoint and a second waypoint from the representation of the road, determines a linear projection of the source point to a line connecting the first waypoint and the second waypoint, determines a first estimate of a longitudinal component of the source point in a road-based reference frame based on the linear projection, the first estimate being on a curve representing the road between the first waypoint and the second waypoint, determines a second estimate of the longitudinal component from the first estimate, determines a coordinate of the source point in the road-based reference frame from the second estimate and operates the vehicle with respect to the object using the coordinate.

Growing community of inventors

Herzliya, Israel

Roi Reshef