Optimal Path Planning of a Mobile Robot using Quadrant Based Random Particle Optimization Method

The Random Particle Optimization (RPO) method, used for determining the optimal path from a specified initial pose of a mobile robot in a static or dynamic environment, models obstacles and target as repulsive and attractive Coulomb potentials respectively. It is a sensor-based method drawing its methodology from bacterial foraging just as Genetic Algorithms (GA) are based on Darwin’s genetic evolution of species. The RPO, like GA and random search methods such as the Particle Swarm optimization (PSO) and Ant Colony Optimization (ACO) are effective in obtaining global optima for large Non-Polynomial NP-hard path-planning environments. It has the advantage however that it is not restricted to ‘link points’ which are usually incorporated in GA algorithms. This study aims to investigate a computational acceleration scheme in RPO for obtaining the optimal path for deterministic NP time hard problems. The use of bacterial foraging, or ‘artificial points’ on the next time step of a mobile robot are randomly biased towards the quadrant of interest, in line with the Euclidean path of shortest distance in a 2D workspace, thus reducing the computational processing. Thus, the sample size is reduced with the advantage of less on-line processing for an FPGA-based autonomous robot. The resulting optimal path is compared with the standard RPO as well as an arbitrary-swing RPO to show the computational advantage.