Motion planning of a robot corresponds to computing a sequence of actions to find a path from an initial state to a goal, where the mission is accomplished. A planning algorithm searches the state space of the robot, which is all the states that the robot can be in, to compute these actions. However, this procedure has high computational complexity, such that only an approximate representation of the state space can be used. This approximation can be obtained using a set of pre-computed motions, named motion primitives, that dictate the transitions between states. Therefore, the success of a planner in finding a solution and that this solution is good is related to the selected motion primitives. However, in most cases this choice is arbitrary. This proposed research will answer the question: how to select a good set of motion primitives so that they will be effective in solving planning problems.

2021

2024