Research On Control Method Of Unmanned Ground Vehicle Based On Predictive Control

As the core of intelligent transportation network,unmanned ground vehicle(UGV)has great significance in military and civilian applications.Because of the complex dynamic characteristics,strong nonlinearity,underactuated characteristics,the control algorithm of UGV has always been the focus of academic circles.In this paper,the obstacle-avoidance control of UGV,the steering control with tire blowout and the distributed control scheme of multiple UGVs are studied.Firstly,a two-layer control structure based on model predictive control is designed for obstacle avoidance of UGV.In consideration of the actual control input constraints,the planning layer is responsible for reconstructing trajectory by choosing different planning algorithms according to different obstacle types to avoid obstacles.When the obstacles are static obstacles,the continuous trajectory planning problem is discretized into a discrete nonlinear optimization problem by using the direct collocation method;when the obstacles are dynamic obstacles,the planning layer uses the rolling horizon control algorithm to plan the trajectory in real time to adapt to the movement of the obstacles.The tracking layer is responsible for tracking the planned trajectory quickly and accurately.A continuous-time model predictive control(CTMPC)algorithm is introduced and improved for the under-actuated characteristics of the UGV.And an extended state observer is designed to observe the disturbances in the system online.Secondly,a continuous-time adaptive model predictive control method is proposed for the steering control problem of UGV with tire burst.Tire burst will the increase of tire rolling resistance coefficient and the decrease of cornering stiffness,which will easily cause the vehicle to deviate from the lane or even rollover.The parameter changes caused by tire burst are transformed into uncertainties,and an adaptive fuzzy observer is designed for on-line observation.On this basis,the vehicle dynamic model is linearized by feedback linearization,and the analytic control law of continuous predictive control is designed with Taylor expansion and control input saturation constraints.Finally,distributed controllers based on nonlinear model predictive control(NMPC)are designed for the formation control and formation conversion tasks of multiple UGVs.The collision avoidance constraints between multiple UGVs are considered with trajectory estimation errors,and the terminal constraints are designed to guarantee the stability based on the minimum invariant set theory.With the movement of the time window in the control algorithm,a local feedback control law is designed to estimate the UGV trajectory in the future.At the same time,the packet loss data can be estimated accurately when the multi-vehicle network communication is unstable.