Accurate Optoacoustic and Inertial 3-D Pose Tracking of Moving Objects With Particle Filtering

摘要

In industrial environments, the determination of the positions and orientations of tools in complex manual assembly processes allows an automated monitoring about the fulfillment of the process and a quality control. This contribution presents an optoacoustic indoor localization system based on combined distance and inertial measurements that is able to undertake this positioning task. In doing so, an accurate distance measurement is achieved with ultrasound and infrared. This novel system is able to localize multiple moving objects attached to transmitters simultaneously by measuring the unilateral distances to room-fixed receivers and fusing these measurements with inertial navigation system data (acceleration, angular velocity, and magnetic field) in a particle filter (PF). In doing so, the PF is determining the objects' poses by using the accelerometer and gyroscope data for a pose prediction in a propagation step and by utilizing the distance measurement and magnetometer data for an estimation correction. A commercially available, optical reference system is used to record the dynamics of human hand movements in an assembly scenario in order to analyze these dynamics for a system performance evaluation. Subsequently, the novel, an optoacoustic system is tested by applying these dynamics to an electrical linear axis on which a transmitter is fixed. Compared to the reference system, the median position error achieved is below 2.6 cm and the median orientation error is less than 10.0 degrees, even when going beyond the hand movement dynamics recorded previously.