Research On Key Techniques Of Six-Axis Force/Torque Sensor For Special Robot

Large extravehicular space manipulator is an actuating mechanism which is used for space working. It is the main carrier of on-orbit service and space exploration. Multi-axis force/torque sensor, which is installed on the wrist of the space manipulator, is one of the most important sensors, and also a key component for space robot force control and teleoperation. However, space working is such a special background that brings huge challenges to the design of multi-axis force/torque sensor. On the one hand, space tasks require larger measurement range, higher stiffness, and higher sensibility. On the other hand, the sensor should achieve higher reliability to adapt to the environment of space.In this thesis, a six-axis force/torque sensor, which could be used as a component for the large extravehicular space manipulator, is designed. A sampling system is integrated in the sensor. Then, a static calibration experiment is conducted, followed by studying the decoupling algorithm. At last, the sensor is successfully developed.According to the application background and the technical indicators, an elastic body based on cross-beam is designed. With the help of ANSYS software, the size is optimized by using finite element analysis method. Then, an assembly structure with anti-overloading capability is designed for the packaging and assembling of the sensor.Different from the conventional way, the signal acquisition module is integrated in the sensor so as to accomplish the local digitalization of the signal, which makes it easier to be mounted on the manipulator. In order to increase its reliability and flexibility, a dual channel acquisition circuit is designed, including a high-speed acquisition channel and a low-speed one.To achieve high precision, a novel calibration system is designed to provide large and accurate force/torque source and two decoupling algorithms are proposed. The online decoupling algorithm is based on calculating dynamic decoupling matrices in partitioned space, which cooperates with the low-speed channel, having simple operations and comparatively lower precision. The offline algorithm is based on BP Neural Network, which cooperates with the high-speed channel, having comparatively more complex operations but higher precision.Experimental results show that the designed six-axis force/torque sensor works well with high precision and reliability.