Positioning,Navigation And Planning Of Biped Robot In Complex Environment

Biped robots have remarkable ground adaptability due to their intermittent footground contacts,and can walk stably in the presence of obstacles and gaps as a result of their foot structure.Considering their high mobility and extraordinary environmental adaptability,robotic research all over the world has been specifically focusing on addressing the problem of motion planning and control of robotic systems.The DARPA Robotics Challenge,which is held by the U.S.Department of Defense,shows the cutting-edge biped robot technology and indicates the current research direction in the motion planning and control strategies for biped robots in complex environments.In this paper,motion planning and control of biped robots in complex environments are explored with an aim to improve their motion capabilities in complex environments.Generally large biped robots are equipped with a sophisticated feedback control framework to achieve stable locomotion.However,this is not the case for biped robots,which have difficulty in walking stably because of limitations in the hardware.In this paper,the biped locomotion model and control system are explored and the whole gait planning process of biped robots is presented using the linear inverted pendulum model as theoretical framework.Then a universal feedback control system for gait improvement in biped robots is proposed based on the stability analysis of biped locomotion.Using IMU data,joint angle sensor data and plantar pressure sensor data as inputs and the trajectory of the centroid as feedback,this system obtains stable biped locomotion and reduces biped locomotion stability requirements in hardware design.Afterwards,compared to wheeled robots embedded with synchronous mapping and localization,biped robots fail to accurately control the direction and distance to walk when executing the mapping and localization algorithms,which will cause great interference to the localization algorithm.This paper first builds a visual localization and navigation system for biped robots.Using dense maps based on point cloud map splicing to navigate,the system realizes biped walking path planning using the artificial potential field method,and performs path compliance processing and footprint sequence generation.In order to improve the accuracy of the odometer estimation during the walking process,an odometer estimation method suitable for biped robots is brought about.It first attempts to use the particle filter method to deal with the IMU noise to eliminate the falling impact force on mileage estimation.Aiming at the cumulative errors of the particle filter-based IMU odometer,a multi-sensor odometer estimation fusion method is proposed.The noise parameters of the IMU are estimated in advance,and then grounded on the extended Kalman filter method,the visual localization data are fused with the IMU data to estimate robot mileage information.Experiments have shown that the multi-sensor odometer estimation fusion method can eliminate the accumulated errors caused by the insufficient walking accuracy of biped robots,and improve the odometer estimation accuracy and robot localization accuracy.Subsequently,this paper investigates falling protection motion control for biped robots in case of instability.Biped robots are prone to lose balance and fall down while walking.To minimize damage,this paper puts forward a multi-objective falling trajectory optimization method by analyzing robot falling dynamics and falling protection constraints.This optimization method considers the lower legs,thighs,torso and arms of the biped robot as a multi-step inverted pendulum,and by analyzing the falling motion equations and the stability conditions for the inverted pendulum,it simulates and plans a protective posture during falls.Then combined with both kinematic and physical constraints,it constructs a multi-objective trajectory optimization algorithm for the falling robot,which can optimize each of the joint angle and angular velocity,reduce the angular momentum of the robot during falls,and get the minimum kinetic energy at the point of landing.This method has been able to protect the robot from ground collision and reduce damage to robot hardware.Overall,the method has been proved effective through simulation and prototype experiments.Last but not least,as the biped robot is a non-linear complex system with multiple degrees of freedom,it is impossible to achieve stable locomotion in a complex environment if traditional control methods are applied.This paper examines the trajectory planning for complicated robot locomotion.To improve problems such as complicated foot-end trajectory calculation and discontinuity in position or speed of the steering connection trajectory in present complex locomotion trajectories,trajectory planning methods for stairs climbing,slope climbing,curved path walking and one-step steering are improved using the robot dynamic model,which can simplify geometric calculations for locomotion trajectories and enhance trajectory continuity for walking connection.In order to improve the locomotion stability and flexibility of complex trajectories,based on the hybrid particle swarm evolution algorithm,ZMP constraints and kinetic energy constraints are constructed considering the locomotion stability and trajectory continuity to optimize the trajectories for complex trajectories given the kinematical constraints of biped robots and to improve the ZMP stability margin and dynamic characteristics of walking trajectory during complex motions.Experiments have shown that this method can effectively avoid falling of biped robots in stairs climbing and slope climbing,increase the maximum single-step steering angle of the steering gaits,and improve the locomotion stability and flexibility of robots in a complex environment.In a nutshell,by investigating the localization and complex locomotion planning and control of biped robots in complex environments,this paper aims to plan and optimize the complex motions for biped robots on a theoretical level,and to tackle key technologies related to locomotion in complex environments in coordination with high-precision localization and navigation methods.Further,this paper works on theoretical and methodical approaches to improving complex locomotive abilities of biped robots,hoping to promote the research and development of biped robots in our country to some extent.