| CPC B60W 60/0011 (2020.02) [B60W 30/143 (2013.01); B60W 40/04 (2013.01); B60W 40/06 (2013.01); B60W 60/0015 (2020.02); B60W 60/00274 (2020.02); B60W 2540/18 (2013.01); B60W 2554/80 (2020.02)] | 20 Claims |

|
1. A method of operating a vehicle, comprising:
automatically planning a motion of the vehicle with an automated motion planner Pm, which conducts a search in a motion planning domain Dm to determine and optimum motion plan comprising an action a which solves a motion planning problem specified by an initial pose xi representing a first position and a first orientation of the vehicle, a goal pose xg representing a second position and a second orientation of the vehicle, and a road network providing constraints on the vehicle, and produces a control signal
δ,θ , wherein δ∈Δ is an acceleration and θ∈Θ is a steering angle of control operation sets Δ and Θ, such that δ,θ ←Pm (x, x′) to achieve a collision-free motion trajectory ξ connecting the initial pose xi and the goal pose xg taking into account the constraints on the vehicle, to incrementally advance the vehicle toward a goal, the optimum motion plan comprising action a being selected dependent on an associated utility of the incremental advance that represents a change in value, based on at least a safety of the motion trajectory with respect to an environment of operation of the vehicle, a benefit resulting from the incremental advance, and a cost of the incremental advance; andautomatically planning a task for the vehicle with an automated task planner Pt, the task comprising a sequence of the actions to advance the vehicle toward the goal, selectively dependent on an optimization of an aggregate prospective utility of the task comprising the benefit resulting from achieving the goal, the costs associated with the sequence of actions, and the safety of the sequence of actions to advance the vehicle toward the goal, wherein the planning a task comprises:
calculating a plan p∈P of a set of plans P based on a task planning problem within a task planning domain by Dt, including a set of states S within a factored state space, such that each state s∈S is defined by values of a fixed set of variables, the task planning problem comprising an initial state sinit∈S, a goal state sg∈S, and a set of actions A, each action a∈A being defined by its starting state s and resulting state s′, the plan p consisting of a sequence of transitions represented as p=
s0, a0, . . . , sN-1, aN-1, sN , where s0=sinit, sN=sg;defining a utility function dependent on at least a cost function Cost(s,a,s′) and a safety function Safe(s,a,s′) of conducting action
s,a,s′ in state s; andproducing an optimal plan
![]() based on the utility function.
|