Gait Planning And Stability Control For Biped Robot Based On Preview Control

The ability to walk stably is always an important research field of robotics because it is a necessary premise for a humanoid robot to serve people. Preview control theory is a widely used method to generate the robot’s COM(center of mass) trajectory due to its advantages. For example, the method math model is very simple and it can adjust COM trajectory in advance based on ZMP(zero moment point) trajectory in the future and generate very smooth trajectory. On the other side, the method’s disadvantages are poor real time property and large tracking errors because of its simple math model. In this thesis, the proposed method based on preview control has improved the property of real time and tracking errors and made the robot enable to reject the disturbance. The main research issues and results are summarized as follows.First, the real time property of the preview control method is negative correlated with its preview time. And the tracking error is positive correlated with its preview time. This thesis improves the real time property through decreasing the preview time and compensates the tracking error using multi-links model dynamics compensation. So this thesis solves the contradiction between real time property and tracking error.Second, most of the methods can’t adjust the footstep placement when the robot suffers from disturbance. This thesis proposes anti-disturbance walking controller which can calculate the new footstep placement using inverted pendulum with an inertial flywheel model to make the robot recover from perturbation. In addition, in order to reduce the impact force between the ground and the foot, the thesis proposes stable walking controller to enhance robot’s ability to walk stably.At last, a humanoid robot model based on BHR-5 is built in a dynamics simulation environment using Adams software. The proposed method is validated by simulations and experiments on BHR-5.