Abstract (EN):
Purpose - The purpose of this paper is to characterise the TWIN-RRT* algorithm which solves a motion planning problem in which an agent has multiple possible targets where none of them is compulsory and retrieves feasible, "low cost", asymptotically optimal and probabilistically complete paths. The TWIN-RRT* algorithm solves path planning problems for both holonomic and non-holonomic robots with or without kinematic constraints in a 2D environment. Design/methodology/approach - It was designed to work equally well with higher degree of freedom agents in different applications. It provides a practical implementation of feasible and fast planning, namely where a closed loop is required. Initial and final configurations are allowed to be exactly the same. Findings - The TWIN-RRT* algorithm computes an efficient path for a single agent towards multiple targets where none of them is mandatory. It inherits the low computational cost, probabilistic completeness and asymptotical optimality from RRT*. Research limitations/implications - It uses efficiency as cost function, which can be adjusted to the requirements of any given application. TWIN-RRT also shows compliance with kinematic constraints. Practical implications - The practical application where this work has been used consists of an autonomous mobile robot that picks up golf balls in a driving range. The multiple targets are the golf balls and the optimum path is a requirement to reduce the time and energy to refill as quickly as possible the balls dispensing machine. Originality/value - The new random sampling algorithm - TWIN-RRT* - is able to generate feasible efficient paths towards multiple targets retrieving closed-loop paths starting and finishing at the same configuration.
Idioma:
Inglês
Tipo (Avaliação Docente):
Científica
Nº de páginas:
10