A method for the motion planning of humanoid robots' stepping upstairs based on an improved NSGA-II

Based on the analysis of the certain limitations of current methods for optimization of humanoid robots' motion caused by their optimization of single objectives such as energy, stability and speed, a optimization method based on multi-objective optimization to optimize the motion parameters of a humanoid robot in stepping upstairs was presented. In consideration of the low efficiency of fast nondominated sorting of the NSGA-II, a typical nondominated sorting genetic algorithm (NSGA) with the elitist tactics, an improved NSGA-II method based on self-adjusting binary search trees was proposed, and by using it, the motion parameter optimization for a humanoid robot in stepping upstairs was achieved. The humanoid robot's energy consumption and stability before and after the optimization were measured and compared by computer simulations and experiment. The experimental results show that the use of this method can overcome the disadvantages of the single objective optimization, and effectively realize the humanoid robot's motion planning when it stepping upstairs in the circumstances of meeting multiple objectives requirements.