The Configuration Design And Fault Detection Research For A Class Of Reconfigurable Modular Robots

With the rapid development of modern industry, robots have been applied in manyfields, such as aerospace, military and other areas. Due to be restricted by its mechanicalstructure, traditional robot can be used to do something simple. Besides, parameters ofdesign process are very complicated. Compared with traditioned robot, the reconfigurablemodule robot possesses advantages of low requirements of working conditions and lowcost, so that it can be more used. Therefore, reconfigurable module robot attracted moreattention from scholars at home and abroad.In order to satisfy the complex requirements, and to lengthen its life and improvereliability, the reconfigurable module robot requires to be equipped with the fault detectionsystem, which is dispensible for robots used in the areas of industry, military, agricultureand other areas.In this paper, the configuration design and fault detection research for a class ofreconfigurable modular robots have been studied. The main works are summarized asfollows:1) This paper has discussed the configuration design problems for a class ofreconfigurable module robot. The system of reconfigurable module robot belongs to thesystem of complex multiple inputs and outputs. This system has the characteristics oftime-varying, nonlinear, strong coupling, etc. In this paper, though geometric forms of theNewton-Euler iterative algorithm, the dynamics model of the robot is derived andestablished.2) A variety of fault-tolerant control design has been studied in the robot system. Thispaper mainly studies the multi-sensor information fusion in module and the analysis ofreliability, and presents a fault-tolerant controller design based on H-infinite filter.3) In view of fault detection problems for the reconfigurable module robot, the designmethod based on adaptive H-infinity filter has been put forward. First of all, each part ofreconfigurable robot model can use the rules of T-S fuzzy modeling. Secondly, through thedesign of a filter and two additional weighted matrixes, the residual outputs have beenminimized without faults; On the contrary, the residual outputs have been maximized withfaults. This method can improve the sensitivity of the fault against the residual. Thirdly,due to the influence of the interference to residual output, effects of interferences havebeen minimized. It can make sure the fault can be accurately detected by the system in time. Finally, indexes of the linear matrix inequality (LMI) satisfy each performancesbased on GKYP lemma.4) Two freedom degrees of the reconfigurable module robot systems have been built,and the signals of faults have been detected. The simulation results prove that theproposed fault detection algorithm can detect the sensor faults effectively.