Walking Planning And Control For Slip Prevention Through Humanoid Robot Whole-body Coordination

Encouraging achievements have been made during the last decades in humanoid robot research and great process has been made.Many technical problems remain to be solved for the humanoid robot to assistant human in real applications,among which stable walking is the basic and quite important one.It is easier for the humanoid robots to trip over when they slip during biped walking.During the development of the new robot prototype,we find that translational and rotational slip could occur at the same time,which is always ignored by other researchers.The former studies of slip prevention mainly focus on translational slip prevention or on rotational slip prevention independently.Very few works discuss how to control the humanoid robot to prevent the translational and rotational slip occurrence simultaneously with the natural,human-like walking pattern.Therefore,this dissertation presents walking planning and control for slip prevention through humanoid whole-body coordination.It is supported by the projects “Theory of Humanoid Robot's Multimode Locomotion and Transition” from National Natural Science Foundation of China.The main research issues and results are summarized as follows.Firstly,the motion capture system and reaction measurement system are utilized to collect and analyze the human walking data.From the bionics point of view,the human natural walking and slip prevention mechanism is summed up,which is the bionic foundation for the slip prevention scheme on the humanoid robots.And the overview of the whole walking planning and control scheme is proposed on basis of the human slip prevention mechanism and humanoid robot dynamics characters.Secondly,the walking planning for slip prevention is proposed.The relationship between center of mess(COM)and zero moment point(ZMP)on basis of linear inverted pendulum model(ILPM)is derived.However,the ILPM is simplified and cannot reflect the dynamics characters of the swing leg,which have significant effect on slip during the biped walking.Therefore,an extended three mass model is proposed.And a yaw torque compensation method based on human walking mechanism through upperbody and lowerbody coordination is also proposed.Then the optimal trajectories of the whole body coordination are generated based on the extended three mass model,the ZMP stability margin and slip prevention constrains.Thirdly,the walking controller for slip prevention is proposed.The dynamics characters of three mass model and the quantitative relationship between the increasement of ZMP and the acceleration of COM are presented.And the slip prevention controller based on multi-axis motion is proposed according the above quantitative relationship.The humanoid robot will adjust itself according to ZMP detection and slip trends detection.And also,a spring-damping model is utilized to decrease the impact while the swing leg loading on the ground.Finally,the effectiveness of the proposed method is validated through both simulation and experiments on a real humanoid robot BHR-5.