基于眼到手视觉伺服的移动机器人模型预测控制

Model predictive control of mobile robot in eye-to-hand visual servoing

摘要:

针对具有速度约束的移动机器人视觉轨迹跟踪问题，提出了一种基于LOQO内点法的模型预测控制方法。在眼到手框架下，首先建立了移动机器人误差模型，并对该误差模型进行离散化，给出了移动机器人视觉伺服跟踪的代价函数。同时考虑到实际中移动机器人存在速度约束问题，将代价函数的最小化问题转换为带输入约束的模型预测控制问题。然后采用障碍函数法将移动机器人的速度约束转化为等式约束并采用拉格朗日乘子法引入到代价函数中。进而，利用LOQO内点法求解具有速度约束的最小化问题，得到基于视觉的轨迹跟踪控制器。最后，通过仿真验证了所提算法的有效性和优越性。

Abstract:

A visual trajectory tracking problem is studied for mobile robot with velocity constraint by using model predictive control method incorporating the LOQO interior point method. Firstly, a discrete mobile robot error model is established under the eye-to-hand framework, and a cost function for the visual servo tracking of the mobile robot is given. Considering the existed velocity constraint problem for the mobile robot system, the minimization problem of the given cost function is transformed into a model predictive control problem with input constraints. Then, a barrier function method is used to transform the velocity constraint into equality constraint by using the Lagrange multiplier method. Furthermore, a LOQO interior point method is used to solve the minimization problem with velocity constraints, and a vision-based trajectory tracking controller is obtained. Finally, simulations verify the effectiveness and superiority of the proposed algorithm.