A Stabilization Control Method For The Underactuated Bipedal Walking On Compliant Ground

The underactuated bipedal walking is becoming a hotspot of robotics,because of the high-level human-likeness,low energy-consumption and good maneuverability.However,the ground compliance is the main reason for its instability in the real ground environment.The so-called ground compliance is the ability of deformation of ground pavement under load.The real ground must be compliant.Unfortunately,because of the difficulty in compliant ground modelling and the impossibility in pre-detecting the exact ground compliance coefficients,the traditional stabilization control method based on the "robot-ground" rigid contact assumption is no longer applicable.Thus,making the underactuated bipedal walking system sense and adapt to the ground compliance and walk stably with invariant step length on the compliant ground should be the first challenge of our research.Furthermore,in the real ground environment,the perturbation is always hybrid type,such as the compliance being always accompanied with the local pavement surface depression,which is mainly caused by the fatigue and corrosion.In this case,the robot should lengthen its step size and stride over these depressed zones to stabilize the walking process.The fundamental condition to realize a stable underactuated bipedal walking is the system can arrive at a periodically stable state.However,the varying of step length must broke the existing periodic stability of the walking system,which should damage the walking stability.Thus,resolving the contradiction between the periodicity and aperiodicity of underactuated bipedal walking is the other challenge of our research.To overcome these challenges,in this thesis,by taking a planar point-feet biped robot as the experimental object,the principle for underactuated bipedal walking instability on the compliant ground is studied,an adaptive feedforward control(AFC)strategy used to realize a stable walking with invariable step length is developed,and a high-level control strategy based on AFC for stabilize the underactuated bipedal walking with variable step length is proposed.Firstly,the analysis of the effect of ground compliance on the underactuated bipedal walking stability.In the numerical experiment,the compliant ground was modelled as a spring-damper system equivalently,and according to the documented mechanical properties of compliant materials common in daily life,five stiffness coefficients and seven damping coefficients were applied.In the physical experiment semi-rigid polyvinyl chloride(PVC),and chipboard were used as the pavement of compliant ground and concrete was used as the pavement of rigid ground.Then,with the comparison of numerical and physical experiment results,the principle for underactuated bipedal walking instability on the ground compliance was obtained.Secondly,the development of adaptive feedforward control method.Based on the principle for walking instability and inspired by the human gait characteristic,a stability criterion only concerned with the horizontal velocity of the center of mass(CoM)of robot was proposed;and then,an adaptive feedforward control strategy only based on the robot's CoM states was developed.Through the numerical experiment and physical experiment,the availability and adaptability of AFC was demonstrated and the stable underactuated bipedal walking with invariable step length on the compliant ground was realized.Finally,the development of stabilization control strategy for walking with variable step-length on the discontinuous compliant ground.Inspired by the human stride pattern,by taking the advantage of the inherent characteristic of walking process controlled with AFC,a high-level control strategy for the underactuated bipedal walking with variable step-length was developed.The analytic proof,numerical experiment and physical experiment showed that this high-level control strategy was able to stabilize the underactuated walking with variable step length on the discontinuous compliant ground.