Research On Mechanic Design And Walking Control Of The Small Four-Legged Platform

Unmanned ground platform is one of the hottest focuses in the research of militaryrobot. In comparison with wheel-pedrail walking gear, foot walk is a mode of motion thatbears the characteristics of bionics, and theoretically the most powerful adaptability to anylandform and over-obstacle capacity. Foot walk mainly includes two-legged, four-leggedand six-legged. Four-legged is more stable than two-legged, and less redundant andcomplicated than six-legged in control.As a multi-branch walking gear, small four-legged platform bears time-varyingtopology motion structure, which also makes it a complex non-linear kinematic system withtight coupling and multiple variables. On the basis of new modeling method, a four-leggedplatform kinematic model is established in this paper. Furthermore, forward kinematicsolver is designed through recursion programming principle, and analytical method isapplied in a monopod gear to analyze and solve inverse kinematics. In the meantime, thesolver and visualized display of inverse kinematics which lay a theoretical andprogramming foundation for gait planning are realized.A new motor-driven leg gear is designed with the concept of coupling driven. Thus,this leg gear can satisfy small rotational inertia and big drive capacity. By means ofsimulated analysis, this leg gear is compared with traditional leg gear in the rate of energyconsumption, in which the rationality of this coupling leg gear is further demonstrated.Moreover, a simulation test is conducted on the comparison of six mass distributionplatforms in energy consumption. It is achieved that under crawl gait, although there’s nosignificant difference in energy consumption, the closer the mass becomes to the rotationaxis of motion, the smaller the consumed energy is.The crawl gait of four-legged platform is optimized by adjusting the side direction andexcursion mode of engine gravity center. Moreover, the distance of the forward gravitycenter is equally distributed to the execution of each phase, which effectively improves thestability of continuous crawl gait and reaches the goal of both speed and stability. Thefour-legged platform cannot effectively guarantee its stability in uphill and downhill. Inconsideration of this question, posture feedback theory is put forward, and thecompensatory articular angle is calculated in accordance with deviation of pedal extremities.Thus, how to adapt to topographical changes is solved from the quantitative angle of view and the four-legged platform’s over-obstacle capability and adaptability to any landform issubstantially enhanced.In the study of gait of trot in opposite angles on the four-legged platform, invertedpendulum model, energy orbit principle and “virtual leg” concept are applied in the controlof jogging in opposite angles on the four-legged platform. Thus, an ideal ZMP trail isplanned. The control position and speed of leg joints are calculated through the technique ofcubic spline and demonstrated in webots simulation software.In the study of jumping gait on the four-legged platform, sinusoidal model and fourierseries are applied to design the mathematical model of jumping gait. In consideration of thedifficulty in solving the complicated relation between model parameter and motion, geneticalgorithm is adopted to optimize parameters of the mathematical model of jumping gait todesign fitness function that can evaluate the level of jumping gait. By introducing elitemode to the evolutionary process, the misconvergence of simulation can be avoided.A small principled sample machine of the four-legged platform is developed and aperformance test is conducted on the leg gear driven unit of the principled sample machine.An on-the-spot squatting and rising test is designed to demonstrate the rationality of the leggear design. Next, another test is designed on the rotational gait of the principled samplemachine and crawl gait of the four-legged platform. Thus, the validity of gait planning andcontrol of the four-legged platform is verified.