Abstract:
In this paper, four new methods for finding an optimum path through a grid of different terrains in static condition are proposed. The robot used to follow the path is a novel modular self-reconfigurable robot with 3 Degree of freedom (DOF) called ACMoD. For each type of terrain, the robot can reshape automatically to achieve better performance in terms of time and energy. The performance of the optimization algorithms is evaluated using a pre-defined multi-objective function. These algorithms are: Genetic, Ant Colony, A* and Dijkstra algorithms. In Genetic algorithm, the properties of different terrains are encoded in chromosomes together with the configuration of the robot. In Ant colony, Dijkstra and A* algorithms, the properties of different terrains and the configurations are encoded in edge of graph. The outputs of these algorithms are path and configuration pattern for each block. 3 of these methods are heuristic search methods. The other one, Dijkstra, is a deterministic algorithm. Finally, the accuracy of the path solutions and the convergence rate of the algorithms are compared and analyzed. The simulation results demonstrate that all the proposed methods except Dijkstra not only find the correct solution but also are effective in different situations in term of convergence rate and path. As a result, the path length is reduced successfully and effectively.