With the development of robotics, the power of robot has been improved and the application areas of robot have been enlarged. Robots are required to accomplish complicated taskes in dangerous environments in place of human, such as toxic waste cleanup in a nuclear power station, high-voltage line overhaul, space station restore, ocean exploration and so on. The implement of all this tasks needs a good robot control system that has good architecture and strong functions. Therefore, it's an very meaningful job to do research on good architectured robot controller, as well as to improve the reliability, quickness and flexiblility of robot controller and to extend the application area of robots. This paper is focused on the research on Open Architectured Robot Controller. The content of this paper goes as follows: Firstly, the characteristics of Open Architecture Robot Controller are introduced, researches on the architecture and the implement of the software and hardware of the open architecture robot controller are reviewed, Secondly, a Linux-based Open Architecture Robot Controller that we develped is introduced, the architecture and functions of it are discussed from the hardware components and software modularity. This controller has the function of motion planing, speech control, visual control and remote control. As far as the K10 robot is concerned, the resolution of the singularity of it is discussed, as well as the motion planing algorithm. Also Remote Real-time control is done and the result shows that this system works stably in real time and the motion planing algorithms are correct. Thirdly, the network communication and its implementation of the cross-platform of Open Architecture Robot Controller is introduced, the system architecture, the methods of implementation of the network communication are discussed in detail. Also we discuss the definition of the data format and the implementation of the procedure. Fourthly, visual algorithms are addressed at the high level of the Open Architecture Robot Controller and the representation methods of the contour are discussed in detail. And then a new curve fitting algorithm- polyline splitting algorithm based on chain code- is forwarded. This method can represent the boundary easily and greatly lessen the storage quantity of data on the image. Then remote visual control of robot use the polyline splitting algorithm to lessen the transfer data. Thus, the validity of curve fitting algorithm is confirmed by the experiment.
修改评论