A Hybrid Approach to Motion Planning for Stepping over and Obstacle Avoidance in Humanoids

Obstacle avoidance and path planning of biped are challenging issues addressed by researchers across the globe. Tackling with obstacles in mobile robots generally involves obstacle avoidance by choosing via points away from the obstacle. In addition to this, legged robots particularly humanoids, have the potential to step over or step up on obstacles to cross them. This approach can be more effective in case of certain category of obstacles which are not stretched along the lateral direction and there by not hindering the stance foot placement. This necessitates a separate study of these category of obstacles to tap the full potential of humanoids to step over them where ever possible and thereby reducing the necessity of a path modification. The approach involves per-determining the optimum step length of the robot by energy optimization and dynamic analysis, followed by the selection of best trajectory for stepping over obstacles that are fitted into basic 3-D shapes for easiness of real time computation. A fuzzy based trajectory planning for these basic shapes are studied in detail and the intelligent fuzzy based planer selects the best trajectory, after cross checking the stability of the choice.