Research On Motion Planning And Control Strategy Of Hexapod Robots Walking On Rugged Terrains

Due to the structural redundancy and foothold discreteness of legged robots,they have much superiority in terms of stability and terrain adaptivity.As a result,compared to traditional wheeled and tracked robots,legged systems generally offer better potential to move over complex terrains.Besides,as a kind of typical bio-inspired systems,legged robot can also serve as physical platforms to explore and test those adaptive motions and the corresponding underlying principles exhibited by animals,which helps us to better understand biomimetics,biomechanics and neuroscience.Owing to these potentials and advantages,it is of great importance to study the related techniques of legged robots.As a kind of legged robots,the ones with six legs have received much concern,and to date a number of hexapod robotic systems have been developed.This dissertation also focuses on the hexapod robots,aiming to further enhance their ability to move over complex terrains.With this purpose,the dissertation conducts detailed exploration from three aspects,e.g.,intrinsic properties of the robot,control efforts,and motion planning approach.Reasonable mechanical structure,powerful electronics,and functional sensory devices are fundamental for the robot successfully implementing tasks.First,the hexapod robotic platform used in this dissertation is introduced,and its kinematics is analyzed to offer a reference for the subsequent motion planning and control issues.The motion of a hexapod robot can be roughly divided into two parts.The first part is several legs in support propelling together the torso forward,while the second part is the rest legs swinging in the air based on the torso.Therefore,the intrinsic properties of legs play a direct role on the robot's performance.A good property may be able to improve the self-adaptivity of the robot,thereby reducing the dependence on active control.In order to make better use of those ‘hardware intelligence',the morphology of the robotic legs are analyzed to balance the motion margin of torso in the vertical direction and the peak torque,peak power and energy consumed during walking.Together with the limb sizes and motion range of joints,the optimal leg morphology is obtained.A double mass-spring model is used to investigate the effect of leg distal compliance on impact buffering at touchdown.By comparing the impact magnitude and impact time under various average stiffness and nonlinearities,it is displayed that convex nonlinear compliance behaviors are able to provide better buffering performance by extending the impact time.A simplified joint model is used to investigate the effect of parallel compliance in joints.Actuation of the model is identified from general biological joints and further reduced with a specific focus on muscle elasticity aspect,for the sake of easy implementation.By examining various elasticity scenarios,it is found that concave parallel elasticity is able to achieve a more advantageous balance between energy efficiency and disturbance rejection than linear and convex ones.Effective control is necessary for the hexapod robot to achieve stable and adaptive walking.Based on the distal compliance related analysis,parameters of the leg compliant mechanism are modified.Furthermore,concave distal compliance behavior and concave parallel compliance in joints are achieved through active compensation.To enhance the stability during robot walking,the torso control strategy is proposed and the corresponding foot forces are calculated.Finally,an artificial neural network based adaptive impedance control method is presented to modulate the force at each foot.With this method,impedance parameters are able to be tuned in time through three neural networks.In order to further improve the adaptivity and walking efficiency of the hexapod robot over complex terrains,a whole-body trajectory generation method is explored based on the perceived terrain model and planned motion path.First,a method of terrain feature extraction is proposed to describe the local terrain shape,and a foothold utility function is designed and trained through support vector machine(SVM).On the basis of this,a foothold selection method for the hexapod robot is presented based on the synthesized foothold utility and the support polygon thus formed.Then a supporting movement generation method is proposed to produce reasonable joint trajectories for those legs in stance phase.In the end,the problem of leg swing is formulated as an optimal control procedure that satisfies a series of locomotion task terms while minimizing a biologically-based objective function,which is solved by a Gauss Pseudospectral Method(GPM)based numerical technique.We apply this whole-body trajectory generation method to a simulation robotic platform.Results show that the proposed method is able to successfully generate feasible trajectories over complex terrains.Experimental research is an important part of development of hexapod robots.A series of experimental tests are carried out using the established hexapod robot platform to further analyze and evaluate the effectiveness of the proposed theory and method.The distal compliance experiment is carried out,and the previous theory is verified by measuring the foot force and the compliant deformation in the distal leg segment.The robot body control experiment is carried out to test the effectiveness of the above-mentioned control strategy.The swing movement generation experiment and the support movement generation experiment are carried out to verify the effectiveness of the whole-body trajectory generation method.At the same time,the support movement generation experiment also further validates the feasibility of the proposed foothold identification and selection method.Finally,effectiveness of the adaptive impedance control algorithm and the pose adjustment algorithm for the robot is evaluated by the walking experiment on lightly rugged terrain.The comprehensive utility of the proposed methods is evaluated by the irregular wave terrain walking experiment.The results show that the robot successfully traverses the complex terrain and successfully completes the given walking task.