Deep Learning Prediction of Gait Based on Inertial Measurements
We report the application of recurrent deep learning networks, namely long term short memories (LSTM) for the modeling of gait synchronization of legs using a basic configuration of off-the-shelf inertial measurement units (IMU) providing six acceleration and rotation parameters. The proposed system copes with noisy and missing data due to high sampling rate, before applying the training of LSTM. We report accurate testing results on one experimental subject. This model can be transferred to robotised prostheses and assistive robotics devices in order to achieve quick stabilization and robust transfer of control algorithms to new users.
This work has been partially supported by FEDER funds through MINECO project TIN2017-85827-P.
- 2.Cheng, G., Peddinti, V., Povey, D., Manohar, V., Khudanpur, S., Yan, Y.: An exploration of dropout with LSTMs. In: Proceedings of Interspeech 2017 (2017)Google Scholar
- 3.Gal, Y., Ghahramani, Z.: A theoretically grounded application of dropout in recurrent neural networks. In: Advances in Neural Information Processing Systems, pp. 1019–1027 (2016)Google Scholar
- 5.Kok, M., Hol, J. D., Sch öon, T.B.: Using inertial sensors for position and orientation estimation. arXiv preprint arXiv:1704.06053 (2017)
- 6.Ikehara, T., et al.: Development of closed-fitting-type walking assistance device for legs and evaluation of muscle activity. In: 2011 IEEE International Conference on Rehabilitation Robotics (ICORR), pp. 1–7. IEEE, June 2011Google Scholar
- 15.Zhang, M., Sawchuk, A.A.: A feature selection-based framework for human activity recognition using wearable multimodal sensors. In: Proceedings of the 6th International Conference on Body Area Networks, pp. 92–98. ICST (Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering) (2011)Google Scholar