Active particle in MCL: An evolutionary view

Mobile robot localization is the task of determining a robot´s pose in a known environment, which is one of the most important problems in mobile robotics. The state-of-the-art Monte Carlo Localization (MCL) algorithm requires a large amount of particles and thus converges slowly. Also, MCL performs poorly in low noise sensor input. This paper develops a more efficient algorithm called APMCL. It puts the process of MCL in the CLONAL framework, which is a computational implementation of the clonal selection principle in immune system, and utilizes Particle Swarm Optimization to improve the mutation process. The key idea of APMCL is to take the perceptual model p(y_t | x_t) as the fitness function and then carry out a heuristic searching step on the particles generated in MCL. It drives the particles towards the regions where the value of the desired posterior function is large. As a result, the particles are no longer passively waiting for being filtered in Sequential Monte Carlo process. In contrast, they will actively adapt themselves to much more valuable pose (with higher fitness value) by cognitive study (i.e. mutation) and swarm study (PSO). Experimental results show that APMCL reduces the number of particles needed in conventional MCL and improves the performance of the algorithm regarding to the computation time as well as the localization precision.