Method and system for deterministic trajectory selection based on uncertainty estimation for an autonomous agent
AssigneeGatik AI Inc.
InventorsGautam Narang, Apeksha Kumavat, Arjun Narang, Kinh Tieu, Michael Smart, Marko Ilievski
Life of this patent
Estimating the term…
Abstract
A system for deterministic trajectory selection based on uncertainty estimation includes a set of one or more computing systems. A method for deterministic trajectory selection includes receiving a set of inputs; determining a set of outputs; determining uncertainty parameters associated with any or all of the set of inputs and/or any or all of the set of outputs; and evaluating the uncertainty parameters and optionally triggering a process and/or action in response.
Claims20
A method for operating an autonomous agent, the method comprising:
- collecting a set of inputs;
- determining a set of multiple candidate trajectories based on the set of inputs;
- determining a set of multiple uncertainties associated with the set of multiple candidate trajectories and the set of inputs;
- comparing the set of multiple uncertainties with a set of satisfaction criteria, comprising:
- in an event that at least one of the set of multiple uncertainties satisfies the set of criteria, selecting a trajectory from the set of multiple candidate trajectories based on the comparison;
- in an event that the none of the set of multiple uncertainties trajectories satisfies the set of criteria, triggering an emergency planner to determine a set of actions;
- operating the autonomous agent based on one of the selected trajectory and the set of actions.
A system for operating an autonomous agent, the system comprising:
- a computing system, wherein the computing system collects a set of inputs;
- an out-of-distribution detector in communication with the computing system, wherein the out-of-distribution detector determines a first uncertainty associated with the set of inputs;
- a classical motion planner in communication with the computing system, wherein the classical motion planner:
- determines a first trajectory;
- a set of learned models executable by the computing system, wherein the computing system, with the set of learned models:
- determines a second trajectory;
- determines a second uncertainty associated with the second trajectory;
- a gatekeeper module in communication with the computing system, wherein the gatekeeper module:
- compares the first and second uncertainties with a set of satisfaction criteria;
- and
- an emergency planner in communication with the gatekeeper module, wherein in an event that neither of the first and second uncertainties satisfies the satisfaction criteria, the emergency planner:
- determines a set of actions for the autonomous agent, wherein the computing system operates the autonomous agent based on the set of actions.
Dependencies are inferred from the claim text.