Path Finding Using PSO Cooperated with Randomized Noise Functions

Path planning is a crucial navigation technology for routing and shortest path problems. The paper discusses a modified approach to the collision-free searching problem using particle swarm optimization incorporated with noise functions like Gaussian and Perlin noise. Aiming at PSO’s shortcoming of quickly diving into local minima, the added random noise functions escape the local minima in the convergence process; hence, look for global convergence maintaining fast speed in the early phase. The random sampling improves the particle update procedure to look for broader search space, which is otherwise constraint to global best of population. This variation guarantees solution space exploitation. Particle position vector precision is improved by adding noise (by some predefined factor), and the PSO algorithm is run to get the best particle as a candidate solution. Particle swarm optimization is a low-overhead and easy to implement the technique. Obstacles are incorporated into the algorithm to improve effectiveness. Particles need to reach the destination without colliding with any of the obstacles. Finally, simulation scenarios demonstrate effectiveness considering multiple target positions.