Research On The Body State Estimation Method Of Bionic Legged Robot Based On Multi-Sensor Fusion

State estimation is the most fundamental and critical aspect of research on bionic legged robots.Due to their unique body structure,legged robots have a higher degree of complexity in their movements compared to traditional wheeled robots.Legged robots control the swing of a single leg by rotating joint motors,which in turn drive joint linkages.The joint motors work together to achieve the robot’s forward and turning movements.This mode of movement makes it challenging to estimate the body state information of bionic legged robots.To reliably estimate the body state information of quadruped robots,this paper proposes a flexible state estimation framework that can be used to estimate the body state information of quadruped robots operating in complex real-world scenarios.The proposed state estimation method is mainly based on error state Kalman filtering,which fuses IMU sensor data and legged odometer information to achieve the optimal estimation of the robot’s body state using a multisensor fusion approach.The main research results of this paper are as follows:(1)A D-H parameter model is established based on the joint parameters of a single leg.A forward kinematics model for a single leg is built to calculate the position of the end effector of a single leg in the body coordinate system.Then,based on the transformation relationship between the single leg coordinate system and the body center of mass coordinate system,the transformation relationship between the position of the end effector of a single leg in the body center of mass coordinate system and the joint angles is derived.A single leg inverse kinematics model is also established based on the joint angles.(2)The body coordinate system,IMU coordinate system,and ground inertial coordinate system are established for the quadruped robot,and the robot’s body state variables are defined.The velocity update equation of the IMU in the ground inertial coordinate system is established,and the estimated body velocity information of the IMU is obtained.Based on the kinematic and dynamic information of the body,the legged odometer model is derived.A foot-ground contact detection method that does not depend on force sensors is proposed,which outputs the reliable contact probability between the foot and the ground and incorporates the foot-ground contact probability as a weight into the body velocity estimation to minimize the estimation error of the body velocity.(3)The body state information obtained by IMU inertial navigation and the state information estimated by the legged odometer are fused through the error state Kalman filter,and the error state prediction equation and update equation are established.The optimal error parameters are estimated using the error state Kalman filter equation to correct the IMU drift error and obtain the optimal body state estimation result.Finally,a simulation experiment is designed,and a simulation environment is built using the Gazebo simulation engine in the ROS system.The robot model is controlled to move in a straight line at different speeds in different terrains.The simulation experiment outputs the optimal state information estimated by the fusion algorithm.The body state information estimated using IMU inertial navigation,legged odometer,and fusion strategy is analyzed and compared,and it is finally verified that the proposed fusion estimation framework can effectively suppress IMU drift,reduce the peak noise of legged odometer,and has higher stability and accuracy.