Robust trajectory planning using data-driven reachability corridors

A method to apply data driven reachability corridors in vehicle trajectory planning, comprising:

• determining a path for a host vehicle to take;
• calculating an ideal path incorporating a baseline trajectory from an initial host vehicle start time t0 to a completion of a path change at a final time tf;
• calculating multiple feasible trajectories of the host vehicle;
• defining a guidance tube around the ideal path incorporating the multiple feasible trajectories at multiple points around the ideal path considering multiple distances to proximate vehicles;
• performing an online real-time calculation by a computer to identify multiple baseline goal sets;
• and
• forming a reach corridor within which the host vehicle dynamically reaches a desired target set of the multiple baseline goal sets including wind and local friction variations and working from the final time tf back to the host vehicle start time t0 defining a host vehicle start state.

A method to apply data driven reachability corridors in vehicle trajectory planning, comprising:

• calculating an ideal path for a host vehicle incorporating a baseline trajectory in a host vehicle computer;
• loading an output of a trajectory and motion planning module including a reference trajectory and an operation corridor into a disturbance and reachability refiner module;
• applying data-driven reachable set computations by learning set mappings offline using supervised learning;
• applying an approximating function online, applying the reference trajectory and the operation corridor to compute a reachability corridor for the host vehicle to remain inside of during a maneuver of the host vehicle;
• and
• updating the reachability corridor while constraining the host vehicle computer based on local disturbances using the disturbance and reachability refiner module including changing road conditions, inclement weather and roadway disturbances.