The Locomotion And Control Method Of The Cheetah Robot For Ultar High Speed Running

The locomotion characteristics of flexible, stable, and non-continuous support make the quadruped robot capable on many tasks which wheeled robots could not be done, such as the unknown unstructured environments and excellent ability to adapt to agile motion characteristics. The quadruped robot research is always a hot topic in biomimetic robot research. However, within the further research on the biomimetic robot, the quadruped robot research branches to different ways, adaptability and speed. Further because the large force and the short contact time, the highly nonlinear and the fast switching, make the higher requirements and challenges to the controller. Research on this direction will improve the bionic analysis, robot dynamics, and many other aspects of the research. The work on this area could have a good application in the military, industrial, life and other aspects.In this paper, we make cheetah as the research object. Through physiological and morphological analysis, the robot design was absorbed into a large number of bionic elements. Using antagonistic multi-joint legs model, we built a musculoskeletal cheetah quadruped robot system. The research on the ultra high speed running departed into two parts: One, using a simplified model with compliant legs, researches and analysis the use of hip torque and angle adjustment to control the speed and torso attitude; Other, utilizing the multi-joint leg model with muscle characters, investigates the nerve stimulation on muscles and joint flexible driven to develop the control method of multi-joint locomotion. Combining the research of body movement and leg movement control, a bionic control strategy was built for the cheetah robot high speed running. The simulation analysis and the experimental tests showed the validity of the control and the feasibility of robotic system design.Based on the analysis of the characteristics of biological motion running, a principal simplified model is established. Utilizing the Poincare map and Newton-Raphson method, a lot of stable fixed point was found, and base on the analysis of these points, we found that the energy adjusting is the main source of the velocity changing. Thus, a high speed control based on energy adjustment was established. The method achieved a good performance on speed variation and great acceleration. On this basis, through non-symmetrical adjustment of the energy, it achieved the robot posture control and half bound gait running.For the velocity feedback problem, this paper proposed a bio-inspired speed feedback algorithm, by a large number of the simulation statistical analysis and the research of biological principles on vestibular reflexes. In this method, a relationship map was established between pitch movement and forward movement. It achieved the robot could sense its movement all by its proprioceptors. Further, the simulation results showed that the method could improve the smoothness of the robot movementOn the other hand, in order to obtain excellent locomotion performance, this paper constructed a musculoskeletal structure robot. In order to achieve the similar biological characters on leg movement, toe trajectory, ground reaction force, and joint compliance, two control strategy were proposed to implement on the stance phase and flight phase: the muscles control in the stance phase based on neural mechanism; the combining control method for the antagonistic joint movement. The method showed good performance on leg motion.Finally, the running control method was proposed under the above researches, which constructed with a similar biological neural control structure. The simulation analysis of virtual prototype verified the validity of the control method. The robot obtained the similar leg swing trajectories and ground reaction force profile with animals, and gained a robust capability on perturbance. Utilizing the pneumatic muscle, the cheetah robot system was established with the musculoskeletal structure. The experiment with the real robot validates the design and the development of the structure and the controller.