Research On Robrot Programming By Demonstration With On-line Obstacle Avoidance

In recent years, with the development of space technology and the increasing application of spacecrafts in diverse fields, the space manipulators increasingly show their significant advantages in finishing the on-orbit maintenance tasks and the space robot technology receives more and more attention among countries. However, due to the particularity and complexity of the space environment, the space manipulators should be easier and more convenient to operate with highly safety. Based on the above operation requirements, in this paper, a study on the programming by demonstration method and the online obstacle avoidance method is made to provide certain theoretical support for the development of the space robot technology in our country. In this paper, the research content is mainly divided into the following parts:Firstly, in order to realize the control and operation of the space manipulators, in this paper, it builds a kinematics model of manipulators and makes an analysis of the manipulators’ singular configuration. It creates a teleoperation platform for the on-orbit maintenance system based on the virtual reality technology, which can realize the forecast simulation of manipulators and gives operators a realistic vision telepresence as well as provides a visual basis for the design and simulation of the Preprogrammed trajectory. At the same, it sets up communication between the teleoperation platform and other parts of the system, which completes the interface design of the teleoperation platform.Secondly, in order to reduce the operators’ programming pressure and skill requirements, in this paper, it comes up with a programming by demonstration method based on the Least Square Support Vector Machine. This programming by demonstration method, focusing on the immovable obstacles in the working environment of manipulators, can guarantee the completion of tasks by overcoming the disturbance of position and time within a certain scope. And this programming by demonstration method carries on a modeling of the manipulators’ tip trajectory through the dynamical system; what’s more, by including the Least Square Support Vector Machine, this programming by demonstration method makes full use of the existing kinematics examples to put the dynamical system into practice. Compared with the previous programming by demonstration method, this programming by demonstration method, with better generalization ability, can regulate the properties of the generated trajectory by changing the parameters of the Support Vector Machine. In this paper, it conducts a simulation through this programming by demonstration method, which proves to be feasible, and it also analyzes the impacts of the relevant parameters of Least Square Support Vector Machine on generating trajectory.Thirdly, in order to dealing with the suddenly appearing obstacles in the manipulators’ working environment so as to guarantee the safety of manipulators, this paper introduces into the above programming by demonstration method an online obstacle avoidance method based on the dynamical system, which can be applied to the trajectory described by the dynamic system. In this online obstacle avoidance method, a dynamic system modulation matrix is created to regulate the trajectory of the dynamic system in line with the size, position and shape of obstacles, which can avoid obstacles by decreasing the speed in the normal direction of the current position along obstacles decrease to zero when manipulators move to the edge of obstacles. This method only makes bigger changes to the trajectory near obstacles with other parts remaining the same. In this paper, there carries on a simulation of the online obstacle avoidance method by matlab to prove its feasibility.Finally, in this paper, by applying those two methods mentioned above to the on-orbit maintenance system, the positioning electric tools experiment and the catching a ball with obstacle avoidance experiment are finished, which physically proves the reliability of the system and the feasibility of the algorithm.