Inertial navigation for divers

Abstract

A SCUBA diver would like to know his/her position underwater, rela- tive to the dive start position. GPS is not an option as the signals do not travel underwater. Transponder/pinger systems have already been developed, but these are expensive and require configuration by the boat operator and the diver.
A simple standalone affordable dead-reckoning system is needed, but it is inherently difficult to determine the diver’s movements. The system needs to be very cheap, so that price is not a barrier to its deployment. The system needs to be small.
The Study Group investigated the possibility of such a system making use of cheaply available accelerometer, gyroscope and magnetic field sensors, similar to those included in many modern smart phones.
The Study Group captured data using an Android device strapped to a skateboard to simulate the type of movements a diver might make underwater. Different modes of movement were evident from filtered versions of the sensor outputs, so indicating that a pedometry based solution ought to be feasible. The Study Group then formulated and investigated the feasibility of a generic dead-reckoning system. Although sensors provide more data than is strictly necessary, significant errors arise from imperfect calibration and from noise for which the Study Group derived estimates of the resulting drift in position over time. The accuracy of practical numerical integration schemes in the context of rotating frames was investigated, and a Kalman filter was used to reduce error in the orientation data by combining accelerometer and gyroscopic data.