Real-Time Vehicle Localization Using Steering Wheel Angle in Urban Cities

Abstract

Whether it is a small autonomous shuttle picking up and dropping off passengers, a robot navigating a large warehouse, a pizza delivery autonomous vehicle, or a truck fleet delivering goods and services, the precise localization of a moving object plays an important role in its safety and reliability. An integrated system composed of an onboard Inertial Measurement Unit (IMU) and a Global Positioning System (GPS) is utilized in vehicles nowadays and can precisely determine the location of a vehicle in real time; however, the vehicle localization accuracy degrades significantly even during the short duration of unavailable satellite signal. In this work, we propose using steering wheel angle and odometer data to determine the vehicle location during GPS satellite outages. Several test-driving experiments were conducted using an OBD-II Vehicle Interface (VI) and a tablet. We augmented our approach using reference GPS coordinates to enhance the vehicle location and rectify bias caused by odometer readings. We compared our approach with the vehicle’s GPS navigation system in an urban environment and verified that our proposed approach performs better. Comparing our approach to IMU has shown that the former predicted locations more accurately and with fewer error drifts.