An Incremental Fuzzy Algorithm for the Balance of Humanoid Robots

Abstract

Figure 11 shows the controller’s performance and response for an x-direction impulse. The response was fast, approximately 2 seconds, until the controller reached a ZMP value near to zero. This feature allows the biped robot to gain stability even during walking (figure 13), maintaining the ZMP always inside the support polygon. Our experiments with the fuzzy PD incremental controller algorithm demonstrated that it is computationally economic, all running in a PIC microcontroller, and appropriated for balance control. The algorithm was successfully used in real-time with the biped robot “Dany walker” (videos available at http://www.inf.fu-berlin.de/~zaldivar). The algorithm proposed in this chapter could be also used in other robots structures with a different dynamic, and even with other degrees of freedom. It would be only necessary to adjust the controller’s gain parameters to the particular structure. We plan to use the information given by an inclinometer along with the ZMP value. In this case, the goal of the bipedal balance robot control will be to achieve an inclination value of cero and to maintain the ZMP at the center, or inside of the support polygon. The bipedal robot used in this work is part of a project that is being developed at the Freie Universität Berlin.