No abstract

No abstract

No abstract

This paper presents a practical approach for a mobile robot, to track a reference trajectory with the nonlinear model predictive control scheme with collision minimum error and without collision. Most controllers avoidance which is implemented on a mobile robot with two including PID control, doesn't include future state of the differential wheels. In model predictive control, also called plant to calculate the current control input. But the Model receding horizon control, cost function is formulated to Predictive Control (MPG) predicts the future state using open minimize tracking error. The optimal control input is solving a .. .. discrete nonlinear optimization problem over a pre-described loop plant dynamics for a prediction horizon. Initally, the prediction horizon based on a gradient descent method. Input MPC has been widely used for controlling chemical process and state constraints are implemented using a penalty function. plants, but its applications are expanding to robot control and The implemented controller minimizes the cost function to large scale distributed control problems [4] [5] [6] [7]. through on-line optimization, making it possible to avoid Although the MPC strategy has been widely implemented on obstacles with a natural and flexible trajectory. The tracking a . . . performance and the obstacle avoidance ability are verified aevarie of appl eas, itwsailt manayssuissill through the realistic simulation. remains main problem[8]. However, many successful implementations have been proposed in the last two decades[9][10]. Principally, the features that set MPC apart I. INTRODUCTION from other control laws are its on-line optimization and Recently, applications involving robots have been constraints. The cost function of the MPC optimization expanding from simple industrial work to field work, with a problem can be formulated with the state and control variety of robots being developed for exploration, constraints of the system to control, the MPC can reflect the surveillance and even military applications[l][2][3]. In object of control for the system properly. particular, many researches are focusing on methods to allow Many researches have been performed on formation control multiple robots to track a reference trajectory without problems, in which multiple robots track their reference collision with obstacles or other robots. The Korea Institute trajectory. In the leader-follower approach, to generate a of Science and Technology (KIST) has been developing a formation for a group of robots, the reference trajectory is network based field robot, that is capable of surveillance and computed relative to the trajectory of the leader [11][12] [13].. . t s h i operation. In the process, But the collision avoidance problem still remains main reexlration,avebeenperf to as thetrajmilitaryackin problem. In this paper, we designed a tracking controller researches have been performed on trajectory tracking XXXĉ ontrol, collision avoidance, and formation control. In order based on MPG generates optimal con...

No abstract