Return to food for thought Return to food for thought
The question of whether robots will one day replace humans is not new and is highlighted whenever a new invention leads to job losses by performing the tasks it is designed for more efficiently and at a lower cost than the people it is believed to replace. Bio-inspired underwater vehicles equipped with artificial intelligence (AI) are advancing rapidly, offering new capabilities for exploring and monitoring aquatic environments. These vehicles, designed to mimic the locomotion, adaptability, and navigation abilities of marine creatures, can access areas that are challenging and potentially hazardous for divers. They can also perform repetitive or long-duration tasks without fatigue, which is a significant advantage over human capabilities. As an example of such machines already available to operators, we can refer to the “underwater swimming manipulator” concepts such as the one provided by Eelume, which has been previously described in an article called “About Underwater Swimming Manipulators” in the section “Equipment to Take into Consideration” on our website.
However, while this article provides an overview of the capabilities of these new-generation machines, it is advisable to consult the studies on this topic in the "Studies of ROV and AUV Concepts" section of our database or other relevant sources for a broader perspective. Given the wide range of studies available, it is wise to select a few that illustrate concepts such as "target detection and exploitation", "automatic path finding and selection", "manipulation", and "general design and propulsion".  "Target detection and exploitation" functions encompass abilities such as detecting, classifying, and enumerating devices, algae, fish, etc. The following documents, among others, may be taken as references to understand this point: • "A survey of AI techniques for control of underwater vehicles" by Pepijn van de Ven, Colin Flanagan, & Daniel Toal. • "Deep Learning on Underwater Marine Object Detection: A Survey" by MD Moniruzzaman, Syed Mohammed, Shamsul Islam, Mohammed Bennamoun, and Paul Lavery. • "Robust Underwater Object Detection with Autonomous Underwater Vehicle: A Comprehensive Study" by Dipta Gomes & Dip Nandi (American International University-Bangladesh) • "Underwater target recognition methods based on the framework of deep learning: A survey" by Bowen Teng and Hongjian Zhao. • "Deep learning with self-supervision and uncertainty regularization to count fish in underwater images" by Penny Tarling, Mauricio Cantor, Albert Clapés, Sergio Escalera. • "Automated Detection, Classification, and Counting of Fish in Fish Passages with Deep Learning" by Vishnu Kandimalla, Matt Richard, Frank Smith, Jean Quirion, Luis Torgo, and Chris Whidden. • "An Overview of Underwater Vision Enhancement: From Traditional Methods to Recent Deep Learning" by Kai Hu, Chenghang Weng, Yanwen Zhang, Junlan Jin, and Qingfeng Xia. • "Deep learning based deep-sea automatic image enhancement and animal species classification" by Vanesa Lopez-Vazquez, Jose Manuel Lopez-Guede, Damianos Chatzievangelou, and Jacopo Aguzzi.
"Automatic pathfinding and selection" is a function that enables an underwater vehicle to autonomously select its path to a target and return to its docking station or launching vessel. The following documents, among others, provide insight into the ongoing research concerning this aspect of artificial intelligence: • “Learning Dynamics and Trajectory optimization for octopus inspired Soft Robotic Manipulators”, by Thomas George, Thuruthel Egidio, Falotico, Federico Renda, and Cecilia Laschi. • “Adaptive Navigation Algorithm with Deep Learning for Autonomous Underwater Vehicle”, by Hui Ma, Xiaokai Mu, and Bo He. • “A digital twins enabled underwater intelligent internet vehicle path planning system via reinforcement learning and edge computing”, by Jiachen Yang, Meng Xi, Jiabao Wen, Yang Li, and Houbing Herbert Song • A Novel Bio-Inspired Path Planning for Autonomous Underwater Vehicle for Search and Tracing of Underwater Target, by Adnan Elahi Khan Khalil, Shahzad Anwar, Ghassan Husnain, Atif Elahi, Zhang Dong. • An Overview of Machine Learning Techniques in Local Path Planning for Autonomous Underwater Vehicles, by Chinonso E. Okereke, Mohd Murtadha Mohamad, Nur Haliza Abdul Wahab, Olakunle Elijah, Abdulaziz Al-Nahari, and S.Zaleha.
"Manipulation" refers to functions that enable remotely operated and autonomous underwater vehicles to grasp objects with full control, even if they are fragile and require careful handling. Manipulation also includes operating various mechanisms without damaging them. The following studies, among others, can be considered references for this topic: • “Soft Manipulators and Grippers: A Review”, by Josie Hughes, Utku Culha, Fabio Giardina, Fabian Guenther, Andre Rosendo, and Fumiya Iida. • “Control Strategies for Soft Robotic Manipulators: A Survey”, by Thomas George Thuruthel, Yasmin Ansari, Egidio Falotico, and Cecilia Laschi • “Model Based Reinforcement Learning for Closed Loop Dynamic Control of Soft Robotic Manipulators”, by Thomas George Thuruthel, Egidio Falotico, Federico Renda, and Cecilia Laschi. • “Learning from Northern clingfish (Gobiesox maeandricus): bioinspired suction cups attach to rough surfaces”, by Petra Ditsche, and Adam Summers. • “Research development of soft manipulator: A review”, by Lu Zongxing, Li Wanxin, and Zhang Liping. • Actuation Technologies for Soft Robot Grippers and Manipulators: A Review”, by Shadab Zaidi, Martina Maselli, Cecilia Laschi, & Matteo Cianchetti. • “Applications of Bioinspired Reversible Dry and Wet Adhesives: A Review”, by Minsu Kang, Kahyun Sun, Minho Seong , Insol Hwang, Hyejin Jang, Seongjin Park, Geonjun Choi, Sang-Hyeon Lee, Jaeil Kim and Hoon Eui Jeong. • “Bioinspired Underwater Adhesion to Rough Substrates by Cavity Collapse of Cupped Microstructures”, by Yue Wang, and René Hensel. • “Underwater Robot Manipulation: Advances, Challenges and Prospective Ventures” by Sara Aldhaheri, Giulia De Masi, `Eric Pairet, and Paola Ardon.
"General design and propulsion" refers to concepts that mimic the forms, propulsion methods, and behaviours of underwater creatures. This includes machines designed with bio-inspired exoskeletons that enable access to areas unreachable by conventional machines and mitigate collision-induced accelerations, thereby reducing stress on the vehicle's components and minimizing environmental impact. "General design and propulsion" also encompasses a new generation of software to control machines capable of performing tasks autonomously or designed to assist the operator efficiently. The selection below may be considered to enhance understanding of this future aspect of underwater vehicles: • “A hybrid dynamic model for bio-inspired robots with soft appendages - Application to a bio-inspired flexible flapping-wing micro air vehicle”, by Mathieu Porez, Frederic Boyer, and Ayman Belkhiri. • “Improved Lighthill fish swimming model for bio-inspired robots - Modeling, computational aspects, and experimental comparisons”, by Mathieu Porez, Frederic Boyer, and Auke Ijspeert. • “SAUV - A Bio-Inspired Soft-Robotic Autonomous Underwater Vehicle” by Fabian Plum, Susanna Labisch, and Jan-Henning Dirks. • “Motion control of unmanned underwater vehicles via deep imitation reinforcement learning algorithm” by Zhenzhong Chu, Bo Sun, Daqi Zhu, Mingjun Zhang, Chaomin Luo. • “A Perspective on Cephalopods Mimicry and Bioinspired Technologies toward Proprioceptive Autonomous Soft Robots” by Goffredo Giordano,Marco Carlotti, and Barbara Mazzolai. • “Underwater Soft Robotics: A Review of Bioinspiration in Design, Actuation, Modeling, and Control” by Samuel M. Youssef, MennaAllah Soliman, Mahmood A. Saleh, Mostafa A. Mousa, Mahmoud Elsamanty, and Ahmed G. Radwan. • “Recent Progress in Modeling and Control of Bio-Inspired Fish Robots”, by Boai Sun, Weikun Li, Zhangyuan Wang, Yunpeng Zhu, Qu He, Xinyan Guan, Guangmin Dai, Dehan Yuan, Ang Li, Weicheng Cui, and Dixia Fan. • “Advances in Autonomous Underwater Robotics Based on Machine Learning” by Antoni Burguera, and Francisco Bonin-Font. • “Reinforcement Learning for Autonomous Underwater Vehicles via Data-Informed Domain Randomization” by WenjieLu, Kai Cheng, and Manman Hu.
After reviewing these documents, one might conclude that the use of divers will soon become obsolete, given that robots can perform all the tasks typically assigned to divers without experiencing fatigue or requiring decompression, and with greater accuracy. It is true that divers from my generation, particularly those who began their careers in the eighties and nineties, have witnessed many tasks traditionally performed by divers being gradually transferred to various underwater vehicles. Therefore, it is reasonable to anticipate that the changes brought about by the introduction of robotics in underwater operations will continue. However, contrary to what some might believe, I do not think this signifies the end of the era for divers. Rather than replacing divers outright, AI-equipped bio-inspired vehicles should serve as tools that complement human efforts. For example, ROVs and AUVs are commonly used to undertake preliminary surveys, monitor environments, and perform data collection tasks, which can reduce the risks and physical demands on divers. Additionally, even though AI and robotics technologies are advancing, there are still limitations in terms of decision-making under unpredictable conditions, handling unexpected scenarios, and performing tasks that require high skills and critical thinking. Thus, these marvelous machines are usually specialized for particular tasks and cannot handle situations beyond those their designers have planned. It is important to consider that there are humans behind artificial intelligence, whatever the software's complexity and the impressive results it may provide. Thus, there will always be situations where human skills, decision-making capabilities, and adaptability are crucial, particularly when nuanced judgment is necessary. Another point to consider is that the deployment of such technology involves economic considerations such as the cost of development, deployment, and maintenance. Therefore, assuming that bio-inspired underwater robots will replace divers is an inaccurate analysis that has often been made previously. For instance, during the 1990s, Norwegian authorities responsible for oilfield exploitation contemplated replacing divers with ROVs. However, they soon realized these machines had limitations and that this decision led to unexpected ethical considerations. Consequently, they reverted to developing the NORSOK standards U 100 to better regulate diving activities rather than eliminating them. Although technology has significantly advanced in favour of robotics over the past 30 years, the core issue remains unchanged. Thus, the appropriate approach is not to eliminate the role of divers but to utilize these technologies alongside divers to enhance the efficiency and safety of underwater operations. To support this reasoning, we can refer to the space industry, which, despite extensive use of robotics, has continued to send people to Skylab and plans to send people to the moon and Mars in the near future. Considering the cost and necessary precautions to send people into space, we can infer that doing so is based on substantial reasons.
However, it is clear that the nature of the tasks performed by divers will have to evolve, requiring a change in their professional profile. Therefore, in addition to being highly specialized in specific tasks, they will have to learn to work in symbiosis with these tools, which implies a minimum understanding of how these machines are designed. In addition, divers and supervisors must be adequately equipped to work with these machines and to replace them when necessary or more appropriate. On this last point, companies often show a lack of imagination and investment. Efforts should therefore be stepped up. • "Diver monitoring systems", as described in the article published in the "Equipment to Consider" section, have been on the market for about ten years. Although their use is mandatory in some countries, such as Norway, under the NORSOK standards, they remain underutilized despite the valuable information these systems provide. These devices have the potential to provide more data than they currently do. However, it is clear that manufacturers will not develop a product that is not selling as well as expected.
• Although digital communication systems have advanced and offer improved clarity, their use in operations where heliox and trimix mixtures are breathed is not yet systematic.
• Helmets are essential components of diving equipment that, despite some improvements, remain similar to those used in the 1960s. In contrast, helmets used by military jet and helicopter pilots, such as the “Thales Systems Scorpion”, are equipped with head-mounted devices that provide enhanced vision, target designation, real-time video display, navigation information, and other critical data about the aircraft's major components. Furthermore, modern helmets provided to infantrymen are also equipped with heads-up displays, offering soldiers information such as navigation data, real-time inputs from sensors or drones, night vision capabilities, and secured digital communication. Similarly, heads-up displays that provide GPS maps and other vital information are now available for many models of motorcycle helmets at a reasonable cost. Conversely, commercial divers, who often operate in hazardous environments, are typically equipped with helmets lacking these advanced features, thus maintaining working conditions reminiscent of those faced by pioneers in the 19th century. Technologies such as sonar, which provides clear vision in zero- visibility waters where lighting is ineffective, as discussed in a previous article, have not yet been fully adopted by companies.
Helmet “Thales Scorpion” - Enhanced vision, - Target designation, - Real-time video display, - Navigation information, - and other critical data about the aircraft's major components. (https://www.thalesgroup.com/)
Diver’s helmet ?
Diver’s helmet with integrated Sonar by Composite-Beat Engel ( https://www.composite-be.com)
In conclusion, without the integration of the above technologies, which are easily adaptable to diving conditions, divers will not be able to work in symbiosis with current and future generations of underwater vehicles. Nor will they be able to efficiently replace these vehicles in operations where the use of a diver should be considered preferable. In addition, divers are disadvantaged by the adverse effects of policies and standards that are misinterpreted by managers due to their lack of knowledge, often resulting in panic when a diver is deployed, leading to the issuance of inappropriate policies. Efforts should also be made to correct this last problem. It is what we try to do with our documents that are based on scientific facts and common sence.
Will bio-inspired underwater vehicles with artificial intelligence replace divers?
Click on the octopus to return to the top of the page