Niche Pseudo-Parallel Genetic Algorithms for Path Optimization of Autonomous Mobile Robot - A Specific Application of TSP

Path planning of autonomous mobile robot is pivotal technique for machine intelligence, which aims to find a non-collision path from initial position to objective position according to evaluation functions in an obstacle space[1]. It can be described as traveler salesman problem (TSP), a typical combination optimization problem, which belongs to the wellknown NP-hard optimization[2]. The mathematical definition can be regarded as a map G = (V, E), where each line e∈E has a nonnegative power ω(e). The aim is to find out a Hamilton circle noted with C of map G in order to obtain a minimum power W(C)=∑e∈E (C)ω(e). Some traditional methods such as greed arithmetic, vicinage arithmetic and dynamic programming algorithm[3] do not behave a good performance on combination explosion aroused by rapid increment in exponent within a solution set of mathematic model, also known as the very quick accretion in both aspects of space and time complication along with the increase of degrees. A very promising direction is the genetic algorithm (GA) except for the traditional methods. Genetic algorithm is numerical optimization method[4] based on the theory of genetics and natural selection. It is a generally probabilistic and adaptable concept for problem solving, especially suitable for solving difficulty optimization and evolution problems, where traditional methods are less efficient. An advanced genetic algorithm, niche pseudo-parallel genetic algorithm (NPPGA) is presented based on simple genetic algorithm (SGA), niche genetic algorithm (NGA) and parallel genetic algorithm (PGA) to further improve the GA for robot path optimization. Research on NPPGA is available for lots of practical problems such as path routing optimization, nets organization, job distribution, scheduling optimization etc.