Unsupervised Human Activity Recognition Learning for Disassembly Tasks

In this research work, an unsupervised human activity recognition framework for industrial disassembly tasks has been proposed using Variational Autoencoders (VAE) and Hidden Markov Models (HMM). The framework was developed to recognize sequential human activities from industrial video data without relying heavily on manually labeled datasets. The proposed system focuses on improving activity understanding in human-robot collaboration environments commonly found in remanufacturing industries. The Variational Autoencoder was used for learning meaningful latent feature representations from unlabeled video sequences, while the Hidden Markov Model was used for temporal activity segmentation and sequential activity recognition. The combination of deep representation learning and probabilistic temporal modeling improved the overall recognition performance of the framework. Experimental analysis demonstrated that the proposed system achieved reliable activity recognition accuracy during hard disk drive disassembly operations. The developed framework also reduced the dependency on supervised learning approaches, which generally require extensive manual data annotation and large labeled datasets. This makes the proposed system more scalable and practical for real industrial environments where collecting labeled activity data is difficult and time consuming. Overall, the proposed framework provides an effective solution for intelligent industrial activity recognition and has strong potential for future implementation in smart manufacturing and Industry 4.0 applications. The developed system can contribute toward improving productivity, automation and coordination in future human-robot collaborative industrial environments.

1.                Kingma, D. P., and Welling, M., “Auto-Encoding Variational Bayes,” International Conference on Learning Representations (ICLR), 2014. 2.                Rabiner, L. R., “A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition,” Proceedings of the IEEE, vol. 77, no. 2, pp. 257–286, 1989. 3.                Wang, J., Chen, Y., Hao, S., Peng, X., and Hu, L., “Deep Learning for Sensor-based Activity Recognition: A Survey,” Pattern Recognition Letters, vol. 119, pp. 3–11, 2019. 4.                Aggarwal, J. K., and Ryoo, M. S., “Human Activity Analysis: A Review,” ACM Computing Surveys, vol. 43, no. 3, pp. 1–43, 2011. 5.                Simonyan, K., and Zisserman, A., “Two-Stream Convolutional Networks for Action Recognition in Videos,” Advances in Neural Information Processing Systems (NIPS), 2014. 6.                Carreira, J., and Zisserman, A., “Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 7.                Donahue, J., Hendricks, L. A., Guadarrama, S., et al., “Long-Term Recurrent Convolutional Networks for Visual Recognition and Description,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015. 8.                Srivastava, N., Mansimov, E., and Salakhutdinov, R., “Unsupervised Learning of Video Representations using LSTMs,” International Conference on Machine Learning (ICML), 2015. 9.                Ji, S., Xu, W., Yang, M., and Yu, K., “3D Convolutional Neural Networks for Human Action Recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 1, pp. 221–231, 2013. 10.              Koppula, H. S., Gupta, R., and Saxena, A., “Learning Human Activities and Object Affordances from RGB-D Videos,” International Journal of Robotics Research (IJRR), vol. 32, no. 8, pp. 951–970, 2013.