paint-brush
Real-Time Anomaly Detection in Underwater Gliders: Conclusion and Referencesby@oceanography
1,301 reads
1,301 reads

Real-Time Anomaly Detection in Underwater Gliders: Conclusion and References

tldt arrow

Too Long; Didn't Read

Anomaly detection algorithm is capable of detecting anomalies like remora attachment and shark hit in diverse real-world deployments. Algorithm achieves real-time estimation through a model-based framework by continuously updating estimates based on ongoing deployment feedback. Future work will enhance estimation accuracy by incorporating large amount of glider data into a n data-driven framework.
featured image - Real-Time Anomaly Detection in Underwater Gliders: Conclusion and References
Oceanography: Everything You Need to Study the Ocean HackerNoon profile picture

Authors:

(1) Ruochu Yang;

(2) Chad Lembke;

(3) Fumin Zhang;

(4) Catherine Edwards.

Abstract and Intro

Anomaly Detection Algorithm

Experimental Evaluation

Conclusion and References

IV. CONCLUSION

In this paper, we apply an anomaly detection algorithm to four real glider missions supported by the Skidaway Institute of Oceanography in the University of Georgia and the University of South Florida. On one side of generality, the algorithm is capable of detecting anomalies like remora attachment and shark hit in diverse real-world deployments based on high-resolution DBD data. On the other side of realtime performance, we simulate the online detection on subsetted SBD data. It utilizes generic data of glider trajectory and heading angle to estimate glider speed and flow speed. Anomalies can be identified by comparing the estimated glider speed with the normal speed range. False alarms can be minimized by comparing the algorithm-estimated flow speed with the glider-estimated flow speed. The algorithm achieves real-time estimation through a model-based framework by continuously updating estimates based on ongoing deployment feedback. Future work will enhance estimation accuracy by incorporating large amount of glider data into a


Fig. 15: Comparison of glider-estimated and algorithmestimated W-E (u, upper) and N-S (v, lower) flow velocities for the 2023 USF-Sam deployment based on real-time SBD data

Fig. 16: Comparison of estimated glider speed (red) and normal speed range (green) for the 2023 USF-Sam deployment based on real-time SBD data.
data-driven framework. It is also worth taking into account the impact of the anomaly on the estimated flow speed, aiding in the process of determining false alarms.

REFERENCES

[1] F. Zhang, D. M. Fratantoni, D. A. Paley, J. M. Lund, and N. E. Leonard, “Control of coordinated patterns for ocean sampling,” International Journal of Control, vol. 80, no. 7, pp. 1186–1199, 2007.


[2] D. A. Paley, F. Zhang, and N. E. Leonard, “Cooperative control for ocean sampling: The glider coordinated control system,” IEEE Transactions on Control Systems Technology, vol. 16, no. 4, pp. 735– 744, 2008.


[3] M. Hou, S. Cho, H. Zhou, C. R. Edwards, and F. Zhang, “Bounded cost path planning for underwater vehicles assisted by a time-invariant partitioned flow field model,” Frontiers in Robotics and AI, vol. 8, 2021.


Fig. 17: Ground truth for the 2021 USF-Gansett deployment.


[4] J. Nicholson and A. Healey, “The present state of autonomous underwater vehicle (auv) applications and technologies,” Marine Technology Society Journal, vol. 42, no. 1, pp. 44–51, 2008.


[5] O. Schofield, J. Kohut, D. Aragon, L. Creed, J. Graver, C. Haldeman, J. Kerfoot, H. Roarty, C. Jones, D. Webb et al., “Slocum gliders: Robust and ready,” Journal of Field Robotics, vol. 24, no. 6, pp. 473– 485, 2007.


[6] M. J. Stanway, B. Kieft, T. Hoover, B. Hobson, D. Klimov, J. Erickson, B. Y. Raanan, D. A. Ebert, and J. Bellingham, “White shark strike on a long-range auv in monterey bay,” in OCEANS 2015-Genova. IEEE, 2015, pp. 1–7.


[7] J. E. McCosker and R. N. Lea, “White shark attacks upon humans in california and oregon, 1993-2003,” Proceedings-California Academy Of Sciences, vol. 57, no. 12/24, p. 479, 2006.


[8] L. G. Baehr, “Swimming with sharks: An underwater robot learns how to track great whites,” Oceanus, vol. 50, no. 2, pp. 42–49, 2013.


[9] J. J. Gertler, Fault detection and diagnosis in engineering systems. CRC press, 2017.


[10] J. Chen and R. J. Patton, Robust model-based fault diagnosis for dynamic systems. Springer Science & Business Media, 2012, vol. 3.


[11] K. Aslansefat, G. Latif-Shabgahi, and M. Kamarlouei, “A strategy for reliability evaluation and fault diagnosis of autonomous underwater gliding robot based on its fault tree,” International Journal of Advances in Science Engineering and Technology, vol. 2, no. 4, pp. 83–89, 2014.


[12] X. Wang, “Active fault tolerant control for unmanned underwater vehicle with sensor faults,” IEEE Transactions on Instrumentation and Measurement, vol. 69, no. 12, pp. 9485–9495, 2020.


[13] B. Y. Raanan, J. Bellingham, Y. Zhang, B. Kieft, M. J. Stanway, R. McEwen, and B. Hobson, “A real-time vertical plane flight anomaly detection system for a long range autonomous underwater vehicle,” in OCEANS 2015 - MTS/IEEE Washington, 2015, pp. 1–6.


[14] E. Anderlini, C. A. Harris, G. Salavasidis, A. Lorenzo, A. B. Phillips, and G. Thomas, “Autonomous detection of the loss of a wing for underwater gliders,” in 2020 IEEE/OES Autonomous Underwater Vehicles Symposium (AUV). IEEE, 2020, pp. 1–6.

Fig. 18: Post-recovery evidence of shark strike on USFGansett
[15] G. Fagogenis, V. De Carolis, and D. M. Lane, “Online fault detection and model adaptation for underwater vehicles in the case of thruster failures,” in 2016 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2016, pp. 2625–2630.


[16] Y.-s. Sun, X.-r. Ran, Y.-m. Li, G.-c. Zhang, and Y.-h. Zhang, “Thruster fault diagnosis method based on gaussian particle filter for autonomous underwater vehicles,” International Journal of Naval Architecture and Ocean Engineering, vol. 8, no. 3, pp. 243–251, 2016.


[17] A. Caiti, F. Di Corato, F. Fabiani, D. Fenucci, S. Grechi, and F. Pacini, “Enhancing autonomy: Fault detection, identification and optimal reaction for over—actuated auvs,” in OCEANS 2015-Genova. IEEE, 2015, pp. 1–6.


[18] V. Asalapuram, I. Khan, and K. Rao, “A novel architecture for condition based machinery health monitoring on marine vessels using deep learning and edge computing,” in 2019 IEEE International Symposium on Measurement and Control in Robotics (ISMCR). IEEE, 2019, pp. C1–3.


[19] P. Wu, C. A. Harris, G. Salavasidis, I. Kamarudzaman, A. B. Phillips, G. Thomas, and E. Anderlini, “Anomaly detection and fault diagnostics for underwater gliders using deep learning,” in OCEANS 2021: San Diego–Porto. IEEE, 2021, pp. 1–6.


[20] Z. Bedja-Johnson, P. Wu, D. Grande, and E. Anderlini, “Smart anomaly detection for slocum underwater gliders with a variational autoencoder with long short-term memory networks,” Applied Ocean Research, vol. 120, p. 103030, 2022.


[21] V. Chandola, A. Banerjee, and V. Kumar, “Anomaly detection: A survey,” ACM computing surveys (CSUR), vol. 41, no. 3, pp. 1–58, 2009.


[22] W. Yassin, N. I. Udzir, Z. Muda, and M. N. Sulaiman, “Anomaly-based intrusion detection through k-means clustering and naives bayes classification,” International Conference on Computing and Informatics, p. 49, 2013.


[23] K. Szwaykowska and F. Zhang, “Controlled lagrangian particle tracking: error growth under feedback control,” IEEE Transactions on Control Systems Technology, vol. 26, no. 3, pp. 874–889, 2017.


Fig. 19: Comparison of the estimated (blue) and true (red) trajectory for the 2021 USF-Gansett deployment. The four green circles are the four basis functions covering the whole trajectory.

Fig. 20: CLLE (m) for the 2021 USF-Gansett deployment.


[24] S. Cho, F. Zhang, and C. R. Edwards, “Learning and detecting abnormal speed of marine robots,” International Journal of Advanced Robotic Systems, vol. 18, no. 2, p. 1729881421999268, 2021.


[25] R. Yang, M. Hou, C. Lembke, C. Edwards, and F. Zhang, “Anomaly detection of underwater gliders verified by deployment data,” in 2023 IEEE Underwater Technology (UT). IEEE, 2023, pp. 1–10.


[26] Y. Ruochu, M. Hou, C. Lembke, C. Edwards, and F. Zhang, “Real-time autonomous glider navigation software,” arXiv preprint arXiv:2304.13096, 2023.


[27] X. Liang, W. Wu, D. Chang, and F. Zhang, “Real-time modelling of tidal current for navigating underwater glider sensing networks,” Procedia Computer Science, vol. 10, pp. 1121–1126, 2012.


[28] O. Schofield, J. Kohut, D. Aragon, L. Creed, J. Graver, C. Haldeman, J. Kerfoot, H. Roarty, C. Jones, D. Webb, and S. Glenn, “Slocum gliders: Robust and ready,” Journal of Field Robotics, vol. 24, no. 6, pp. 473–485, 2007.

Fig. 21: Comparison of glider-estimated and algorithmestimated W-E (u, upper) and N-S (v, lower) flow velocities for the 2021 USF-Gansett deployment.
Fig. 22: Comparison of estimated glider speed (red) and normal speed range (green) for the 2021 USF-Gansett deployment.

Fig. 23: ground truth for the 2023 USF-Stella deployment.


Fig. 24: Broken support rails of USF-Stella.


Fig. 25: Comparison of the estimated (blue) and true (red) trajectory for the 2023 USF-Stella deployment. The four green circles are the four basis functions covering the whole trajectory.
Fig. 26: CLLE (m) for the 2023 USF-Stella deployment.

Fig. 27: Comparison of glider-estimated and algorithm-estimated W-E (u, upper) and N-S (v, lower) flow velocities for the 2023 USF-Stella deployment.


Fig. 28: Comparison of estimated glider speed (red) and normal speed range (green) for the 2023 USF-Stella deployment.


This paper is available on arxiv under CC 4.0 license.