Vibration-Based Dead-Reckoning for Vehicle Localization

Abstract

Tracking locations of vehicles such as forklifts in warehouses and factories is crucial since it contributes to the safety of personnel and the output performance of manufacturing and logistics activities. It is necessary to develop a method of locating the vehicles by attaching generic inertial measurement units (IMUs) without any modifications. In this research, we have aimed at providing technologies to estimate locations of the vehicles with the IMUs simply attached to the vehicles based on vibration analysis. We have also developed a novel method of automatic calibration between the proposed vibration signatures and the road conditions without any manual procedures. Integration of acceleration in short terms (within 0.5-2 seconds) and observation of vibration signatures in long terms (> 30 seconds) are fused in the extended Kalman filtering framework to stably estimate the correlation parameters and can also be used to detect changes in the road conditions. In this research, we confirmed that accuracy of localization is below 1-5% of distance travelled on average for electric wheelchairs and handy wagons without any manual calibration procedures by using the smartphones. With aids of generic Bluetooth Low Energy (BLE) beacons, we also confirmed that the error of location can be bounded to below 2 meters on average.