Map Assisted Post-Processing for Looped Running Trajectories Captured by Smartwatches

One common use case for smartwatches is tracking running activities. However, due to size and power constraints, smartwatches cannot be equipped with high-precision positioning modules, resulting in poor real-time positioning output. Running tracks recorded by smartwatches often require post-processing to enhance their quality. Typical post-processing methods include removing outliers, smoothing trajectories, and fusing pedestrian dead reckoning (PDR) and global navigation satellite system (GNSS). However, their ability to improve trajectory quality is inherently limited since all post-processing inputs are solely provided by the watch hardware. Fortunately, the similarity between the smartwatch’s trajectory and the road network provides an opportunity to use the road network as a priori knowledge to improve post-processing. In this paper, we present a new post-processing framework that integrates road network data with smartwatch PDR and GNSS information. To validate the effectiveness of our proposed algorithm, we conducted a series of field tests in real-world environments. The results demonstrate that our method surpasses existing in-house algorithms, achieving an average increase of 20% in positioning accuracy.