Research On Robot Control Strategy For Environments And Input Constraints

This thesis investiaged some kinds of problem usually encountered in robot control systems for the environment and input constraints,such as instability,low precision,bad interaction performance and input saturation.To solve these problems,this thesis adopts adaptive control with other intelligent control methods and employ force observer to estimate the external force.The introduction of visual technology has solved the safety problems brought by manipulating robots in complex environments.Admittance control is a control method applied to the robot-environment control.It achieves good interaction performance by changing the reference trajectory of the robot.To solve the model uncertainties in practical control systems,this thesis use Neural Neworks(NNs)to estimate and compensate the uncertain terms.To satisfy the problem of the control accuracy,this thesis integrates the error transform technique into adaptive NN controller to guarantee both stable and transient phase.The main research contents of this thesis are summarized as:(i)Based on the vision based human motion capture technology,an adaptive controller is designed to study the control problem of Baxter robot in complex environments;(ii)Considering the collision of the manipulator and the external force of the environment,the admittance control method is used to design a robot control system without a force sensor.The adaptive neural network controller integrates the error transformation technology to meet both the transient and steady control requirment.;(iii)Considering the collision of the manipulator and the external force of the environment,an adaptive optimal control method is proposed to achieve the optimal interaction performance;(iv)Considering the existence of external force and input saturation,a neural network adaptive controller is designed to enable the robot to track the reference trajectory with a certain accuracy under the influence of input saturation.Using admittance control,the robot can achieve good interaction performance.