Development Of A Real-Time Pedestrian Localisation System For A Warehouse Environment

Abstract

This thesis documents the design, implementation, and testing of a real-time localisation system prototype that can operate in a warehouse environment. The performance in a warehouse of three radio technologies (Ultra-wide band time of flight measurement, 2.4GHz radio interferometry, and VHF signal strength mapping) were assessed. The ultra-wide band technology performed very well in the tests, and was used for the remainder of the thesis project. It is projected that 19 ultra-wide band radios can cover the tested 10,000m2 warehouse section. Two localisation algorithms were designed under simulation to accomplish the sensor fusion of the radio and inertial motion measurements. The implemented Unscented Kalman Filter achieved better performance than the sequential importance sampling particle filter. The Unscented Kalman Filter was also found to have a lower execution time, making it more feasible to implement on an embedded system. The real-time localisation system prototype was constructed and then tested in a warehouse environment. The prototype achieved localisation within an average of 1.4 metres with 95% confidence under 4.0 metres. This fell outside the initial goal of 2.0 metres at 95% confidence, but the localisation system was still stably tracking the pedestrian with that precision. The electrical components would cost around 30 USD in full production for a system that will last a full 8 hour shift on a single charge.