Research On Measure System Of Move-in-mud Robot Based On Data Fusion

Move-in-mud robot is a new-type and special-use underwater robot, which can perform the hole excavating work to benefit sling passing in the planned trajectory in the mud under water. Hole excavating work to benefit sling passing is an essential procedure for wreck salvaging. At present the task is performed by divers with the mud penetrator, which is laboursome, inefficient and hazardous. Aimed at this problem, with the support of "Basic Technology Research on Move-in-mud Robot Based on Worm Principle" by the National Natural Science Funds, research on move-in-mud robot's measure system based on data fusion is developed in this paper. The measure system is the essential technology in the move-in-mud robot's research, which is also the premise and ensurence for the robot to accomplish the hole excavating work through planned trajectory.The paper reviewed the present situation of the multi-sensor data fusion theory which is a new information integration and process technology. It also summerised the basic method and research aspect of data fusion, analyzed the present situation of the data fusion in robot domain etc.According to the mechanical structure characteristics and given working environment of the robot, the measure project of move-in-mud robot is proposed, based on kinematics analysis of the move-in-mud robot. Aimed at the vermiculation characteristcs of the move-in-mud robot, its location system performs the orientation of the move-in-mud robot by means of relative oriention, using the attitude information and displacement information. Its obstacle avoiding system uses the acceleration sensor to measure whether the robot has fallen across the barrier, which prevents the robot from being destroyed. Its working pressure measure system used several pressure sensors to examine the pressure of the whole air pressure system.According to the measure project, the total measure structure of the move-in-mud robot is designed, which constitutes PC in upper layer, primaryAVR4433 microprocessor in the middle layer and subordinate AVR4433 microprocessor in the underlayer, with serial communication each other. The pressure measure, obstacle avoiding and location systems are designed. Sensors is chosen. The interface circuits and measure softwares are designed.Using the error analysis theory, location error accumulating is analyzed and the LMS arithmetic to reduce the location error of the move-in-mud robot is put forward. The paper also studied the structure of LMS adaptive filter and the theoretical analysis of astringency by deducing the equal square error of the weigh value in the MSN linear filtering. The LMS algorithm is simulated, results of which made clear that this method can effectively reduce the error of the location system.In order to improve the reliability of location system on the move-in-mud robot and reduce the accumulating errors of relative location method, data fusion method based on Kalman filter is advanced. The simulation of this method shows that it can reduce the depth error of the move-in-mud robot. Considering the radiation problem of filtering resulting from the inaccurate model of the Kalman filter, data fusion algorithm based on the fuzzy Kalman filter is advanced. According to the size of the new information from the Kalman filter, the fuzzy controller is used to estimate the weigh coefficient of the filter Kalman plus, and the using rate of new information has been increased. The simulation result shows the fuzzy Kalman filter can effectively reduce the depth error of the orientation system and restrain the the filtering radiation resulting from the appearance of the measure data with big errors.The experiment system of move-in-mud robot principle prototype is set up, performance of the measure system of the robot is validated by the simulation experiment in the lab. The proformance of robot location system in linear trajectory and planar trajectory is validated by the simulation experiment in the lab.