| Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 103 === This study discusses the planning of a three-dimensional global optimization path for robots through the Ant Colony Optimization (ACO) and the Enhanced Ant Colony Optimization (EACO). In ACO, which was first proposed by Dorigo, the transition probability is composed of pheromone concentration and the length of two segments. This study intended to improve the deadlock caused by the cliffs, insufficient exploration rates, and slower convergence rates in the three-dimensional map, while the EACO targeted the correction of two parameters in transition probability: (a) Using the formula of the length of two segments, the weight of the altitude parameter was strengthened; (b) Using the pheromone concentration formula, the local pheromone updating mechanism was strengthened; (c) Using the determination formula, the ramp angle restrictions were added. A comparison of EACO and ACO shows that EACO has the advantages including faster convergence rates, shortening the path by up to 10%, and increasing the exploration rate by 10%. Evidence from computer simulation shows the EACO can accurately derive the best solution (i.e., the shortest global path) in the complex three-dimensional robot path planning environment.
|
|---|
