Real-Time Operating System Based HIT/DLR â…¡ Dexterous Robot Hand Control Platform Design

Dexterous robot hand has become the research hotspot in many universities and research institutes at home and abroad. In the hardware research, dexterous robot hand getting more and more similar to human hand; in the control platform research, the platform that with hierarchical control structure has been widely applied, and proven to have excellent control performance; in the control algorithm research, the application of the modern intelligent control theory has greatly improved the performance of dexterous robot hand, but in-depth study is also needed.In order to study the control algorithm for HIT/DLR II dexterous robot hand, in this paper, we built a real-time control platform by combining QNX real-time operating systems and Simulink technology. It provides an excellent and easy to use environment for control algorithm designing and testing, simplified the research work, and sets the foundation for future research.This paper firstly describes the control platform construct process: in order to achieve excellent control performance, PC and QNX real time operating system are applied as the hardware and software framework for the high-level control platform; in order to achieve the high efficiency of the research and design, we study the control system design method, compared the latest Model-Based Design methods with the traditional ones, and obtain the conclusion that the Model-Based Design methods better than the traditional, with high design efficiency and quality, so Simulink software which with this technology is used as the control algorithm design environment; after appling a series of secondary development, the platform has been constructed by combining the aboves. The platform has integrated control system modeling, simulation, testing, optimization and real-time control into one tool, and shows remarkable advantages in control performance, human-machine interface, good portability and maintainability of software, in comparision with the conventional one.In the rest of paper, we describe the preliminary research work on the platform, that in order to verify the effectiveness of the platform. In order to verify the support for Model-Based Design of the platform, we firstly establish a PWM control model for single finger through system identification, then designed a position controller based on the model, and applied the controller in the experiment, the result shown that the designed controller has almost the same control effect in the simulation and real-time control. To test the effectiveness of platform in complex control, the paper proposes and designs an adaptive grasp algorithm: a continuous search for a stable grasp posture in wide range carried out with a group of ordered gestures, trajectory planning methods and torque locking mechanism. The effectiveness of the algorithm and the platform have verified through a grasp experiment of different texture, size, shape objects.