Research On Remote Robot Control System Based On Video Fusion

Robot technology become one effective ways to extricate from the heavy and dangerous manual labor in 21stcentury, therefore, it is imperative to develop a similar of variety of human intelligent robot in an industrial society to information society. In the nuclear environment, oceans, explosions and fire scene, polar exploration,waste disposal of the harsh environment that can not be closed to, it is imperative that we develop a robot system that have remote operation,intelligent, This system will become an important symbol of human scientific and technological progress. With the visual function is one of the main features of intelligent robots.Traditional teleoperation robot system separate simulation predict and the video feedback. Though the system only with the function of video feedback is able to promptly notify the spot emergency to the operator, lack the predict of the environment, the network delay; Though the system only with the function of simulation predict is able to predict robot trajectory, The operator can control the robot which base on robot trajectory preview, the environment that forecast often serious different from the environment that the actual robot operation, this prediction is one-sided prediction. This system proposed a new method that unifies the video feedback and the simulation predict, which both made up the shortcoming that the virtual environment separate from the reality, and caused the operator like to arrive personally the scene and adjust orders according to the fusion situation to makes further operation. In this way, not only increased the transparency of teleoperation robot system, but also satisfied the accuracy of forecast latency. Regarding the video frequency transmission, H.264 compression standard that adapts to the channel delay strongly is used to compression and decompression the video from the video capture card, thus the video data compression ratio is high and the transmission effect is good. However regarding the simulation robot, 3DS MAX is used to establish the three-dimensional model, so that increased the considerable and the controllable of the model, and completely fused the model and video through establishing conventional coordinates. In view of the inherent delay characteristic on manipulation system, the Markov algorithm is adopt to training time delay, and then the training result is fitted using the neural network, at last, the fitting model is joined in the robot control procedure to forecast the delay.The experiment and the test result indicated that the design of video fusion in telerobot system could satisfy the real-time transmission of the control command. The video data could also fuse smoothly, which has manifested the validity of the delay algorithm.