A metaheuristic optimization algorithm for unsupervised robotic learning

A meta-heuristic search algorithm intended to introduce chaotic dynamics and Levy flights into the algorithm is presented in this paper. Among most evolutionary computation for optimization problem including meta-heuristic search algorithms, the solution is drawn like a moth to a flame and cannot keep away. The fine balance between intensification (exploitation) and diversification (exploration) is very important to the overall efficiency and performance of an algorithm. Too little exploration and too much exploitation could cause the system to be trapped in local optima, which makes it very difficult or even impossible to find the global optimum. The track of chaotic variable can travel ergodically over the whole search space. In general, the chaotic variable has special characters, i.e., ergodicity, pseudo-randomness and irregularity. To enrich the searching behavior and to avoid being trapped into local optimum, chaotic sequence and a chaotic Levy flight are incorporated in the meta-heuristic search for efficiently generating new solutions. The proposed algorithm with quite general objective function is used to study the ability to develop unsupervised robotic learning such as the maze exploring ability.