Time Delay Algorithm Based On Neural Network Of Force Telepresence Technology

Signal time delay between master and slave manipulator is one of the key problems inpractical applications of teleoperation system. For teleoperation system that have forcefeedback function, delay of force interaction signal which are translated from remote area canmake a bad influence on the performance of force telepresence, it not only causeinconvenience to operator, but also do harm in the system’s stability and transparency.In order to achieve an accurate prediction of force feedback signal, characteristics of theteleoperation system and kinetic properties of remote controlled object must be wellunderstand. But remote controlled object is usually unknown, it’s difficult to establish anaccurate mathematical model of the controlled object, sometimes it’s even impossible to dothis work, especially in the condition that remote controlled object usually needs to bereplaced or changed, a lot of energy and work will be spent in the establishment of accuratemathematical model of the controlled object. Currently, neural network can approximate anytype of nonlinear mathematical mapping through automatically studying training sample data.After experiment on the teleoperation system which is with signal time delay, experiment datais dealt to training sample in order to training the neural network, and let the trained neuralnetwork approximate the mapping relationship between system’s motion state variable anddeviation of force feedback signal caused by time delay, is a reasonable solution. In thecondition that time delay exists, RBF neural network can be used to predict whether theremote area is in a force contacting state. If slave manipulator is contact with the environment,BP neural network is used to predict deviation of force feedback signal caused by time delay,and add the deviation signal to the delayed force feedback signal which is transmitted by thecommunication link, so as to reduce the bad influence caused by time delay.Using matlab establish a typical dynamic mathematical model of teleoperation system’scontrolled object to produce force feedback signal, and using program method to add timedelay to the signal. The final simulating result showed that this method can accurately predictthe deviation of force feedback signal caused by time delay, and the curve of predictive signalis very close to the real force feedback signal which is with no time delay. This method dosenot need to add extra subsidiary hardware system, only needs to add motion state signal ofslave manipulator in the communication link, at the same the time technical principle is easyto achieve, can’t bring in extra calculating time delay for micro controller, so it is a goodmethod to deal the force feedback signal time delay problem of teleoperation system.