Global footstep planning with greedy and heuristic optimization guided by velocity for biped robot

Abstract

In order to give full play to the unique movement capabilities of biped robots that are different from traditional mobile robots, and to improve the ability to adapt to the environment, planning an appropriate global footstep sequences is an important way. In this article, we proposed Greedy and Heuristic Quadratic Programming(GH-QP) based on the Quadratic Programming(QP) method to achieve global footsteps for biped robots. Where GH-QP consists of greedy terms, heuristic terms and complementary terms. The heuristic term tries to minimize the number of steps in order to obtain the global optimal solution as quickly as possible. At the same time, we use the reference forward speed of the robot’s as the weight coefficient of the heuristic item to achieve the footsteps which is more in line with the walking trend. The greedy term minimizes the mutation caused by the heuristic term, making the footstep more inclined to the local optimum. The complementary term further enhances the greedy term to reduce the mutation between adjacent steps. We verify the effectiveness and high efficiency of our method through two sets of comparative tests. We experimentally validated our method on BHR-7P biped robot. The footstep sequences planned by our method adapts to the influence of velocity, and exerts the ability of the robot in the continuous planning process.