| As one of swarm intelligent algorithms, Particle Swarm Optimization takes inspiration from natural behaviors of some social creatures. Such novel and promising algorithm is characteristic with fast convergence property and high efficiency in optimizing tasks. Therefore, it has being applied in many fields, including academic and application ones. For swarm robots, they are typical swarm intelligent systems because such systems can be mapped to PSO except for some physical individual properties when they are used to target search. It is obvious that PSO algorithm can be extended to model and control swarm robotic systems. Besides, a series of problems of navigation, obstacle avoidance, and collision avoidance as well as path planning have to be taken into considerations in autonomous swarm robotics. Consequently, all of them have drawn this thesis into research of navigation in generation, and obstacle avoidance in particular.At first, a formal description on characteristics of swarm robotic systems is given to facilitate mapping from PSO algorithm to case of swarm search. Then the author extends PSO to model and control swarm robotic system, focusing on the target search with swarm robots. It is certified that the PSO-style control algorithm can be used to guide individual robots for the searching process by a series of simulation experiments. Besides, path planning in case of swarm search is surveyed, since such problem is inevitable in autonomous robotics. Based on multi-sensor structure and traditional artificial potential method, therefore, a navigation algorithm with obstacle avoidance is proposed in the presence of static and dynamic obstacles as well as absence of full or part of environmental information. Finally, the above mentioned two methods are integrated for the completion of target search task. The simulation results indicate the validity of extended PSO and artificial potential-based path planning algorithm as well as corresponding strategy. |
PDF Full Text Request