Multi-Objective Optimization in Gait Planning of Biped Robot Using Genetic Algorithm and Particle Swarm Optimization Algorithm

This paper deals with multi-objective optimization in gait planning of a 7-dof biped robot ascending and descending some staircases. Both its power consumption as well as dynamic balance margin depends on a few common design parameters. The biped robot should have a maximum dynamic balance margin but at the expense of minimum power. Thus, a conflicting relationship exists between these two objectives. The said gait planning problem has been modeled and solved using two modules of adaptive neuro-fuzzy inference system. The said multiobjective optimization problems have been solved using a genetic algorithm and particle swarm optimization algorithm, separately. Pareto-optimal fronts of solutions have been obtained, which may help a designer to select the most appropriate solution out of several possibilities. Particle swarm optimization algorithm is found to perform better than genetic algorithm, as the former performs both local and global searches simultaneously, whereas the latter is seen to be weak in terms of its local search capability. Therefore, the main contribution of this paper lies with the application of two optimization algorithms to tackle multiobjective optimization in gait planning of biped robot. Keywords—Biped robots; Gait planning; Multi-objective optimization; Pareto-optimal front; Genetic algorithm; Particle swarm optimization.