Gyroscope-enhanced dead reckoning localization system for an intelligent Walker

The paper presents an enhanced dead reckoning system that fuses data coming from odometry and gyroscope for a more reliable position estimate. Taking into account noise and drift properties of both odometry and gyroscope, and employing some rule-based algorithm, the system improves local self-localization by reducing the effect of non-systematic odometric errors caused by unpredictable bumps or uncertainty encountered in the navigation environment. Another type of error which may occur in passive robots has been considered, e.g. lifting up the iWalker and performing some rotational motion while the contact between the rear wheels and the ground is lost. The bias term of the angular rate sensor has been estimated by Extended Kalman Filter every time the walker temporarily stops. The performance of the proposed localization system has been tested on the iWalker platform for both normal and abnormal cases. Experimental results show the efficiency of this low cost dead reckoning system.