Study on Cooperative Control of Flexible Dual Arm Robot

在本项目中，针对柔性双臂机器人在实际工作中的应用，对这类机器人的力与协同控制进行了研究。具体地说，研究了物体的握持、搬运、装配等任务过程的建模，并在此基础上设计了力与协同控制系统。1.在建模方面，提出了一种简单有效的柔性臂集总参数模型，并对其精度和适用性进行了评价。在此模型的基础上，对目前还没有人研究过的柔性臂的可控性进行了分析，发现了不同的不可控构型。2.在力控制方面，将系统建模为受约束的机械系统，并在此基础上进行了分析和控制器的设计。通过实验验证了该方法的有效性。讨论了力控制与振动控制之间的相互影响，阐明了力控制与振动控制之间的相互影响。3.对于协同控制，与力控制的情况一样，将系统建模为受约束的机械系统，并在此基础上设计了协同控制系统。将该控制系统应用于三维柔性双臂机器人，实现了对物体的握持和搬运任务。针对柔性臂的运动学和动力学逆解问题，将其转化为微分-代数方程，提出了一种基于学习控制的求解方法。对解的收敛进行了分析，并给出了一个收敛条件。通过实验和仿真验证了该算法的有效性。

In this project, aiming at application of flexible dual-arm robots to practical tasks, we researched on the force and cooperative control of this type of robots. Specifically, we researched on the modeling of the task process such as holding, carrying and assembling objects, and designed a force and cooperative control system based on the modeling. The results of the project are summarized as follows :1.Regarding the modeling, we proposed a simple but effective lumped-parameter model for the flexible arm and evaluated its precision and applicability. Based on the model, we also analyzed the controllability of the flexible arm that no one has researched and found different uncontrollable configurations that are interesting and importnat.2.Regarding the force control, we formulated the system as a constrained mechanical system, based on which we made analysis and designed a controller. We verified its effectiveness by experiments. We also discussed the interference between force control and vibration control and clarified its geometry.3.Regarding the cooperative control, as in the case of force control, we modeled the system as a constrained mechanical system, based on which we designed a cooperative control system. We applied the control system to a three-dimensional flexible dual-arm robot and implemented a task of holding and carrying an object. The experiment should be one of the world-first.4.Regarding the inverse kinematics and dynamics of flexible arms, we formulated the problem as that of differential-algebraic equations and proposed a new solution to it that is based on learning control. We analyzed the convergence of the solution and derived a condition for it. We clarified its effectiveness by experiment and simulation.