Instrumentation & Measurement Magazine 24-2 - 124

Fig. 6. Pursuer vessel 1 intercepts the evader while vessel 3 surveils it from a distance (Scenario 3).

378, and 289 km away from the IUU ship. The evader is commanded by T::P to follow the same evading strategy as in the
first scenario. The outcome of the pursuit is demonstrated in
Figure 4. The pursuer manage to intercept the evader in 110
min, while the other two vessels draw the region which they
prevent the evader to escape. This simulation shows the flexibility of the proposed RL technique in taking various forms of
missions without having to modify the algorithm structure.
Finally, a third scenario is considered to test the robustness
of the adaptive learning mechanism. To that end, Scenario 2 is
repeated, except that this time, the vessel originally set to intercept the IUU ship (vessel 2) experiences a malfunction (at time
117 min) after intercepting the evader in 110 min, as shown in
Figure 5. At this moment, vessel 1 is commanded to change its
mission from surveilling to intercepting the evader (Figure 6).
It took the new pursuer 213 min to lock down on the evader
from the moment it switched its mission from surveillance to
interception. This is another example on how the proposed decision-making mechanism can easily adapt to mission changes
in real-time with no extra overhead to the operator.

References
[1] " The state of world fisheries and aquaculture: contributing
to food security and nutrition for all, " FAO. (2016) [Online].
Available: HYPERLINK " http://www.fao.org/3/a-i5555e.pdf "
www.fao.org/3/a-i5555e.pdf.
[2] R. Abielmona, " Tackling big data in maritime domain
awareness, " Vanguard, pp. 42-43, 2013.
[3] P. J. B. Sánchez, M. Papaelias, and F. P. G. Marquez, " Autonomous
underwater vehicles: Instrumentation and measurements, " IEEE
Instrum. Meas. Mag., vol. 23, no. 2, pp. 105-114, 2020.
[4] " Larus technologies products (total::insight and
total::perception), " Larus Technologies. (2020) [Online]. Available:
https://www.larus.com/products/.
[5] T. Akinbulire, H. Schwartz, R. Falcon, and R. Abielmona, " A
reinforcement learning approach to tackle illegal, unreported
and unregulated fishing, " in Proc. 2017 IEEE Symp. Series on
Computational Intelligence (SSCI), pp. 1-8, 2017.

Conclusion

[6] D. B. Fogel, " Machine intelligence, " IEEE Instrum. Meas. Mag., vol.

The article presented an online RL decision-making mechanism to tackle the negative impacts of IUU incidents along the
Canadian coastlines. This is accomplished in real-time and
in a completely decentralized fashion where the only feedback needed for each pursuing vessel is its own pose and that
of the evader. No information is required about the dynamic
models of the pursuing vessels or the evader ship. The adaptive learning solution is implemented using a Value Iteration
process and means of neural network approximators. The
solution is shown to be flexible and interactive to dynamic
124

online variations to the pursuer missions. The proposed tool
is being evaluated for the possible commercialization by Larus
Technologies.

9, no. 3, pp. 12-16, 2006.
[7] S. Sun, Y. Yin, X. Wang, and D. Xu, " Robust visual detection and
tracking strategies for autonomous aerial refueling of uavs, " IEEE
Trans. Instrum. Meas., vol. 68, no. 12, pp. 4640-4652, 2019.
[8] M. Gomez Plaza, T. Martinez-Marin, S. Sanchez Prieto, and D.
Meziat Luna, " Integration of cell-mapping and reinforcementlearning techniques for motion planning of car-like robots, " IEEE
Trans. Instrum. Meas., vol. 58, no. 9, pp. 3094-3103, 2009.
[9] P. Chaphekarande and D. Deshpande, " Machine learning
based brain mri estimation method, " in Proc. 2019 2nd Int.

IEEE Instrumentation & Measurement Magazine

April 2021

http://www.fao.org/3/a-i5555e.pdf' http://www.fao.org/3/a-i5555e.pdf https://www.larus.com/products/

Instrumentation & Measurement Magazine 24-2

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 24-2