Development of Autonomous Control Systems of Manipulators

对于机器人的自主操作与控制系统问题，我们得到了如下结果.关节式机械臂自主轨迹规划与控制系统。我们已经开发了一个系统，它是由一个环境教学系统的基础上使用的空间编码方法和几何模型的距离测量技术，算法的碰撞避免和拾取和离开运动规划，和图形运动模拟器。该系统可以自动生成避碰和拣选-分离运动等简单问题所需的基本运动。七自由度机械手的实验研究显示了所开发系统的有效性.高精度的机械手数字控制。为了实现对机器人手臂的高精度控制，提出了重复控制和二自由度控制的理论。设计了橡胶气动执行器的数字控制系统，实验结果表明该方法对高度非线性系统的适用性.用神经网络控制机器人。将Hopfield神经网络应用于机器人逆运动学。该理论可用于关节型和桁架型机器人的自主轨迹规划。针对自行车机器人的轨迹规划问题，提出了一种利用摆线轮廓线约束髋关节和踝关节位置的方法。实验结果表明，有可能实现稳定的行走。此外，一个倒立摆和Hopfield的网络相结合，已提出的自主轨迹生成的机车机器人。仿真研究证明了该方法的有效性。

For the problems of autonomous manipulation and control systems of robots, we have obtained the following results.1. Autonomous trajectory planning and control systems of articulate manipulator. We have developed a system which is composed of an environment teaching system based on a range measurement technique using the spatial coding method and the geometric model, algorithms for collision avoidance and pick-and-departure motion planning, and the graphic motion simulator. Elemental motions required for rather simple problems of collision avoidance and pick-and-departture motion can be autonomously generated by this sytem. Experimental studies for a manipulaotr with 7 d.o.f. showed the effectiveness of the developed system.2. High precision digital control of robot arms. For the high precision control of robot arms, thoery of repetitive control and two degree-of-freedom control has been developed. Also, the digital control system has been designed for rubber pneumatic actuators and the experimental results showed an applicability of the method for highly nonlinear systems.3. Robot control by using neural networks. Hopfield's neural network was applied for the inverse kinematics of robot manipulator. The theory could be used for the autonomous trajectory planning of articulate and truss-structured manipulators. For the trajectory planning of a biped locomotive robot, a method using the cycloid profile which constrains the position of hip and ankle has been developed. Experimental results showed a possibility to attain stable walking. Also, the combination of an inverted pendulum and Hopfield's network has been proposed for the autonomous trajectory generation of biped locomotive robot. The effectivencess of the method was proved by simulation studies.