Mobile App Driven Localization for Motorcycles Based on an Adaptive Unscented Kalman Filter

Precise and robust localization of vehicles on proving grounds and test facilities is an essential requirement for their safe operation. Although motorcyclists are exposed to significantly greater risks of injury compared to drivers of passenger cars and trucks, no such monitoring systems adjusted to motorcycles exist to date. In this work, we present an Adaptive Unscented Kalman Filter (AUKF) based loosely coupled navigation filter for estimating the position, velocity, and orientation of a motorcycle using a smartphone with an external sensor setup consisting of a GNSS sensor and an IMU. A key advantage of the AUKF navigation filter in the context of proving ground monitoring is that it can be used with different vehicles, which have varying vibration characteristics, without manual re-tuning. Adequate parametrization with regard to the GNSS and IMU sensors and the implementation on a suitable hardware system are addressed. The novel implementation of the navigation filter on a smartphone connected to the sensor setup over a wireless interface allows the flexible integration with different devices and sensor systems, which acts jointly with the adaptivity characteristic of the AUKF. Field tests with a motorcycle demonstrate the applicability and robustness of the developed approach, as well as its performance with respect to a total station reference, achieving a mean 2D error of 0.79 m without RTCM correction data. The orientation estimation of the AUKF navigation filter shows qualitatively improved performance and faster yaw angle convergence compared to a regular UKF. In future work, the orientation estimation of the navigation filter will be validated against an attitude and heading reference system, and tested with different sensor setups and motorcycle models.