Numerical Method Of Controllable Cycle Gait Inverse Problem And Controller Design For Bipedal Dynamic Walking Robot

Biped robot has been widespread concern of scholars because it has a humanlike motion mechanism and expands the space for human activities. Biped walking robot has made remarkable achievements in theoretical research and prototype development after years of effort. But it has the following disadvantages, e.g. high energy consumption, low efficiency and unnatural gait, which limit its application in engineering practice. So scholars focus on passive walking principle which was put forward by Mc Geer and opened a new way for biped robot research. However biped robot which is based on the principle of passive walking has a single walking gait and walking environment, meanwhile stability is poor. In order to enrich walking gait and improve stability, contents of this paper are as follows:1) The research object of this paper is the Compass-like robot which is walking on a downward slope passively. Firstly, for the continuous swing phase we use Lagrangian mechanics to make dynamics modeling, then simulation and experimental analysis are made by Matlab software platform. The above work lay the foundation for the future study.2) Using numerical simulation to explore the relationship among the walking gait, the control input and the initial conditions. By entering the virtual gravity moment in the hip and ankle at the same time, the quantitative relational expression of the above-mentioned three is successfully found, which opens a new way for studying inverse problems. Simulation results show that the expression we obtained is right. Then we try another situation, i.e. entering the virtual gravity moment in the hip and ankle respectively and make a qualitative analysis in the end.3) Using mechanical energy derivative method to design a controller based on the reference orbital energy. Mechanical energy is conserved when the robot is in a stable walking process, analytical expression between walking gait and mechanical energy is found by numerical fitting, thereby we design hip controller and ankle controller respectively. Simulation results show that the controller can improve the system’s ability to adapt to the environment and anti-interference ability.4) State-dependent finite time stabilization controller is designed by combining virtual constraint and feedback linearization, which achieves robot’s fully stabilization within one step. Because the angular displacement of supporting leg is monotonic during walking, in order to avoid multi-value phenomenon, supporting leg is selected as the attitude variable and so swinging leg is select as the synchronous attitude variable. Simulation results show that virtual constraint is an effective control strategy for underactuated robots.