Path Planning Based On Modified Artificial Fish Swarm Algorithm

Nowadays, there are a lot of problems need to be optimized. Path planning for mobile robot is one of them, which has the characteristics of nonlinear, complexity, and restriction, it is composed of mechanical, computer, sensor technology and other multi-disciplinary synthesis. Swarm Intelligence Optimization Algorithm is commonly used to deal with this kind of problem based on bionics model, which is made up of multiple individuals, and cooperating in some special way to solve the distribution problem, eventually realized the optimization. Nowadays, Ant Colony Optimization(ACO), Artificial Bee Colony(ABC), and Particle Swarm Optimization(PSO) are widely used. With the development of science and technology, the optimization algorithm is improved; the convergence speed optimization, accuracy and practicability of the requirements are also constantly improving.This paper mainly completed the several jobs.Firstly, this paper analyzes several problems in path planning, which lists several methods currently used more, including the traditional method and the artificial intelligence method, and it points out the advantages and limitations.Secondly, it introduces a kind of swarm intelligence optimization algorithm, Artificial Fish Swarm Algorithm. Principle, algorithm description, parameters, advantages, disadvantages and research status are introduced. The algorithm is a modern heuristic intelligent search algorithm, which is simple, less restriction conditions; the initial parameter sensitivity etc, it has demonstrated its excellent performance and great potential for development in solving complicated problems.Thirdly, according to the existing disadvantages of AFSA, this paper proposes an improved algorithm based on it for robot path planning. The step size is two adjacent raster distances, the food concentration is distance between artificial fish and the target, visual domain is a rectangular matrix. At the same time, congestion factor and vision domain are dynamic adjusted, which ensure the accuracy and convergence speed of algorithm. In this paper, Prey Behavior is default behavior. Swarm Behavior and Following Behavior play a supporting role to accelerate the convergence speed and avoid falling into local optimum. This paper proposes a new parameter-- jump factor, when meets the jump condition, then executes Swarm Behavior and Following Behavior.Finally, the environment model is established by grid method. The feasibility and accessibility of the grid environment has been validated and analyzed in the randomly generated obstacles. After that, the improved AFSA is applied in robot path planning. The simulation results that the algorithm has good environmental adaptability, the convergence rate and optimizing accuracy are improved obviously.