Inertial Navigation system of In-pipe Inspection Robot

Abstract

The main goal of this study is to design and implement a robust inertial navigation system (INS) for in-pipe inspection robot. To achieve this goal, different mechanization approaches, that are derived in different frames or have different implementation methods, are investigated. These methods include INS derived in e-frame, INS derived in n-frame and 3D reduced inertial sensor system (RISS). The INS uses the full inertial measurement unit (IMU) data to calculate the navigation solution, whereas RISS uses encoder, one single-axis gyroscope, and two accelerometers. Advantages and disadvantages are highlighted for each approach. Due to accumulated error in the INS or RISS solution, a sensor fusion based on extended Kalman filter is proposed. The INS is proposed to be fused with encoder's derived velocity with velocity constraints, and with the detected pipe length as measurements to correct INS solution. RISS is proposed to be fused with the detected pipe length only as measurements to correct its solution. Subsequently, the accuracy of the proposed algorithm is verified experimentally. An experimental setup, with a prototype of the in-pipe robot, is designed and built to test and validate our algorithms in a real pipe. The accuracy of the proposed algorithms was around 3 cm after sensor fusion.