Simulation on Improvement of Position Estimation Accuracy in Underwater Using MEMS IMU

In recent years, development of an underwater observation system that is low cost and easy to operate is greatly required with the expansion of demand for marine resources survey acquiring rare metal and so on. The authors developed a free-fall type observation system of which body is glass sphere. Since the glass sphere has durability for high hydrodynamic pressure, retailed camera is able to be set inside the sphere to motion the sea environment. However, the positioning of the observation system is severe problem, because sinking motion is only provided by gravity or tidal force. To realize a low cost measurement system, we try to use MEMS IMU (Micro Electro Mechanical Systems Inertial Measurement Unit) for position estimation. The position by MEMS IMU has sensor noise and caused in large estimation error. In this paper, to reduce the estimation error, we propose the combination of optical range sensor. By adopting light rather than ultrasonic for range sensor, it is easy to adapt to systems using glass spheres. At first, sensor noise is modeled by allan variance calculated from measured stationary data, position estimation simulation using the modeled sensor data was performed. By adding the ranging data, the standard deviation of the position estimation can be reduced in $(\mathbf{x},\ \mathbf{y})= (28.8\%,\ 4.43\%)$ with comparison of estimation by the sensor data of MEMS IMU and the depth sensor.