Accurate Luge Sliding State Measurement Using Asynchronous Low-Quality Multi-Sensor Data

The velocity of the luge is too fast for athletes to feel the state of sliding accurately. The sliding state should be measured accurately through available sensors to help improve athletes’ performance. However, only certain types of sensors usually work at Yanqing National Sliding Center, i.e., inertial measurement unit, ultra-wide band, and airspeed head. Furthermore, the precision and synchronization of all the above sensors are inferior; thus, accurate state estimation cannot be obtained through either single type of sensor. This paper proposes a sensor fusion method using asynchronous low-quality data to make high-precision state estimations based on the Kalman filter. First, we analyze the uncertain time offset of multiple sensors and extend the Kalman filter to interval fusion to fit it. Second, we use finite checkpoints from three photogates and tracks to evaluate the characteristics of luge motion. Third, particle swarm optimization is established to find an optimal offset to generate a state estimation with the lowest cost function. And we speed up the optimization by searching for optimal results from the finite points. Finally, the proposed method is validated with the experimental data at Yanqing National Sliding Center without ground truth and a simulation model incorporating ground truth.