No abstract

Abstract-Accidental falls are crucial causes of death due to injury among the elderly. Many researches about fall detection applied complex algorithms and required heavy equipment. However, these approaches can hardly apply to the elderly's daily life. In this paper, we employ a six-axis gyroscope that integrated in a small smart bracelet. Users who wear the smart bracelet can get the information including acceleration for X, Y, and Z movement, and the rate of rotation in space. Then, we introduce three feature vector generation methods based on the information and feed these three vectors into support vector machine (SVM) algorithm for fall detection. From a dataset of 66 people, we show that the geometric parameters method is the best of the three with a high accuracy (100%), low false alarm rate (0%) and low missing alarm rate (0%) in a simulated home environment.

GPS probe data have been widely used to provide real-time traffic information, and map matching is a fundamental pre-processing step for utilizing the GPS data. Efficient and accurate map matching algorithm plays an essential role in traffic parameter estimation, while most current algorithms focus more on the improvement of map matching accuracy than that of map matching efficiency. Therefore, the paper proposes an ArcGIS-based hybrid topological map-matching algorithm to improve the map matching efficiency with high map matching accuracy. The algorithm performance is testified with filed GPS data. The result shows that the proposed map matching algorithm is feasible for efficient matching with high accuracy in terms of comparatively less number of GPS data.

No abstract

No abstract