Research On Data Fusion Algorithm And Design Of Integrated Navigation System For Indoor Mobile Robots

With the development of science and technology, the research of the intelligent robot has got unprecedented attention. In order to make the robot become truly intelligent, one of the hardest questions we need to work out is autonomous navigation. In the field of indoor navigation, the IMU(Inertial Measurement Unit) work well when they are used to estimate the attitude. However, because of the inherent error, the navigation information will diverge because of the accumulative error when it has been working for long time. The visual sensor has the advantage of wide detection range and it can get rich information. In addition, it will not be affected by electromagnetic disturbance and it will not cause pollution to the external environment. Nonetheless, they are easy to be affected by the change of light and its data updating rate is low because of large computation. WSN(Wireless Sensor Network) navigation system can obtain precise absolute location information, but the ultrasonic wave of the WSN can easily be blocked by another moving object, then the WSN navigation system will be invalid or lose precition. Suppose these three navigation method can be combined, drawing on each other’s strength, then the practicability and robustness of the indoor mobile robot navigation system can be highly improved.In this thesis, using the information of gyroscope, accelerometer and magnetometer from the MEMS device, we can get attitude angles with no cumulative error through complementary filter, and design the visual system as visual odometer, then use the heading information from complementary filter to transform the information of visual sensor for the INS/Visual integrated navigation system. After building the INS/Visual integrated navigation system and the WSN navigation system, in order to prove that single navigation system is not enough to ensure the reliability, some experiments are carried out when the visual information are affected by the change of light and the ultrasonic wave of WSN is blocked respectively. Then through the federated filter, INS/Visual integrated navigation system is integrated with the WSN navigation system to build the federal filtering system. Then add the function of detecting and segregating the subsystem breakdown to the federal filtering system. And then some comparative experiments are carried out to prove that the federal filtering system still work well though the subsystem break down.Through the comparative experiments, the INS/Visual/WSN federal filtering integrated navigation system designed in this thesis can effectively use the absolute location information from the WSN navigation system to correct the relative location error of the INS/Visual integrated navigation system. In addition, because of the function of detecting and segregating subsystem breakdown, the federal filtering system can segregate the faulty subsystem from the federal filtering system in time so as to work safely and stably. Moreover, when the faulty subsystem go back to its normal situation, the federal filtering system can fuse the data of this subsystem into the federal filtering system without delay. These experments has proved that the INS/Visual/WSN federal filtering integrated navigation system designed in this thesis highly enhance the practicability and robustness of the navigation system of the indoor mobile robot.