[time]
2020-12-01T10:37:31+01:00

[team_name]
CorNav

[team_institution]
National University of Defense Technology

[logolink]


[system_name]


[website]


[track]
4

[reference_person]
langping an,mang wang,zheming tu,cahoqun chu,ze chen, shufang zhang

[email]
anlp_gfkd@163.com

[description]
we propose a zero-velocity detection algorithm for pedestrian navigation based on LSTM. Firstly, sensor data sequences under various movement patterns are collected as training samples. After pre-processing, we use the multi-layer LSTM network to train the model of movement pattern recognition and thresholds adaptive adjustment. Secondly, in the process of pedestrian movement, the threshold is adjusted adaptively according to the change of movement patterns, and it is fused with the navigation trajectory solution model in KEF. In addition,We judge the change of height  according on the Data characteristics.

[references]