Dynamic Locomotion And Anti-disturbance Control Of Underactuated Biped Robots

Humanoid robot is the most expected kind of robot to serve people in all kinds of environments. It has always been arousing lots of attention. Although much progress has been made in the humanoid research, there is still a long way to go before the humanoid robots can be used in any practical applications according to the present status. The major drawbacks are the incapability of rapid locomotion and strong anti-disturbance. The underactuated biped robots have no support area and are always under a dynamic stable motion, walking or standing. To break the constraints brought by the traditional stable region methods, this dissertation makes the underactuated biped robots as the research object, and then fully exploits the fast locomotion and dynamic stabilization capabilities of the biped robots. Hopefully, we can explore a new method of dynamic locomotion and anti-disturbance control for humanoid robots.The main contents of this dissertation are:1. Kinematics modeling of the underactuated biped robots based on virtual force control method. Firstly, the underactuated biped robots are classified into two categories, the planar underactuated biped robots and the underactuated humanoid robots. The virtual force control models are modeled separately according to different state phases. The constraints between the virtual forces caused by the underactuated joints are analyzed, too. The relation between the virtual force and the joint torques, as well as the distribution of virtual forces between the legs during double support stance are derived.2. Push recovery for the standing underactuated biped robot using the hip strategy. Kinematic equations are established and solved according to the constraints between virtual forces brought by the Underactuated joints. The prior targets are to maintain constant body height and upright posture, and then the equations of the forward motion is created and solved. The phase portrait of the forward motion is drawn, and a stable orbit is defined to describe the collection of stable motion states. The disturbances are classified into three categories according to the acting positions of the external force/torques. Correspondingly, three kinds of open-loop control are planned using the hip strategy, in order to maintain balance on both horizontal and vertical directions when the disturbances are coming from both ways.3. Research on dynamic walk planning and anti-disturbance control of planar underactuated biped robots. The horizontal motion features druing single stance phase and double stance phase are analyzed. The relation between walking speed and step length/cycle are found according to human gait formula. The corresponding horizontal motion trajectory of the expected speed is planned on a phase portrait. The end velocity of the present step is estimated and the foothold is calculated and modified to accomplish a realtime closed-loop control for horizontal motion. The closed-loop control is anti-disturbance naturally.4. Research on the sprinting motion planning for planar biped robots based on driving coordination between the knee and ankle joints, in order to achieve a rapid locomotion. This only tiptoe landing pattern introduces a redundant degree-of-freedom, which leads to the uncontrollable inner motion of the support leg. To solve this problem, the virtual force of the body and the work output of support leg joints are analyzed, and then the conditions and method of driving coordination between the knee and ankle joints are put forward to maximize the total power of virtual forces. The proposed approach not only solves the uncontrollable inner motion of the support leg motion by balancing the driving load on ankle and knee joint, but also makes full use of the ankle drive capability. The planar bipedal robot with feet achieves a higher speed by the locomotion pattern of sprinting, while the drive capacity of the other leg joints keeps unchanged.5. Research on dynamic walk and anti-disturbance control of underactuated humanoid robots. The forward motion planning of humanoid robots inherits the method of the planar underactuated biped robots. The expected lateral motion trajectory is planned referring to human lateral fluctuation under consideration of the different difficulty in recovering from lateral disturbance in different directions. Realtime modification of the lateral foothold is studied. The anti-disturbance strategies of single step and multiple steps in lateral motion are analyzed along with the influence of disturbance occur time on the difficulty of recovery.Finally, the main content of this dissertation is summarized and key points of future research are discussed.