Digital Twin Emergency Rescue Co-Pilot for India’s Deep-Ocean Human Submersible
The crew inside Matsya6000 during testing.
By Dr. N.Vedachalam • Dr. VBN Jyothi • Dr. R.Balaji
Under the Deep Ocean Mission, a key component of the Blue Economy initiative by the government of India, the Ministry of Earth Sciences-National Institute of Ocean Technology (NIOT) is developing a state-of-the-art, fourth-generation, deep-ocean, battery-powered, scientific, human-occupied submersible: Matsya6000. It is designed to carry three humans down to 6,000-m depth, with an endurance period of 12 hr. and 96 hr. of emergency life support.
The reliability of Matsya’s mission-critical systems is ensured through redundancies. Human-rated configuration for life-critical systems meets IEC 61508 standards. While emergency drop-weights and jettisonable systems ensure safety under extreme loss of buoyancy and entanglements, a drag-anchor-based emergency rescue system will be used to manage the residual risk.
Matsya6000 features: a fully welded titanium alloy exostructure; an 80-mm-thick titanium alloy human cabin; pressure-balanced, oil-filled, lithium-polymer batteries; redundant power, control, communication, and positioning system architecture; a human-rated emergency drop-weight system; rapid localization capability; real-time crew health monitoring; and subsurface hovering capability.
Digital Twin Co-Pilot
The AI-based cognitive digital twin (CDT) co-pilot, Chaitanya, developed by NIOT’s Matsya6000 team and SRM University, Chennai, will support crew in the event of an emergency.
The Chaitanya comprises 14 coupled models of human physiology, engineering equipment, and the ocean environment to enable machine learning (ML) that can predict future system behaviors and suggest optimal actions. Chaitanya is updated with essential sensory information in real time. It simulates predictive scenarios and can support the Matsya crew by generating/redefining protocols that are optimal for survival during 15 specific emergency scenarios. These scenarios include: effective rationing of onboard emergency power to life-critical equipment and oxygen supply for the crew (within and beyond 96 hr.); tracking the ascension of Matsya during positioning system outages; and determining the hovering depth during delayed retrievals.
Chaitanya will comprise four modules that govern Matsya’s emergency rescue system via artificial intelligence/machine learning.
The Four CDT Modules
Chaitanya comprises four modules to regulate power, oxygen supply, navigation and positioning.
The emergency power system module of Chaitanya, Shakthi, is based on a precise mathematical model of lead-acid batteries incorporating machine-learned thermal conditions inside the human cabin at different ocean depths. It provides the emergency operating protocols (EOP) for rationing energy to life-support systems; voltage-sensitive actuators of emergency drop-weights (operating >18 VDC); and jettisoning and emergency rescue systems. Any deviation from the EOP leads to higher energy consumption of battery energy than envisioned and affects the energy availability for other life-support systems. Usage of the underwater acoustic telephone (UAT) on the submersible cannot be restricted during an emergency period, as distressed crew will need to be able to communicate with rescuers. Shakthi predicts the allowable usage rate for UAT during an emergency so that it doesn’t exceed its allocated 8 percent of the energy budget.
The oxygen supply module, Prana, runs on an AI-based crew oxygen consumption model.
The position-track module, Pushar, predicts the likely surfacing position and time, enabling precise positioning of rescue assets for quick recovery of the distressed crew. Pushar works on the dead-reckoning principle, with inputs such as the last known geo-coordinates and machine learning parameters (vehicle buoyancy, salinity profile, ocean current profile, and Matsya’s hydrodynamic behavior). It supports navigation in the event of acoustic positioning systems failure and during emergency ascend scenarios, such as cabin smoke/fire.
The Garuda module determines the optimal hovering depth, considering available propulsion power for station-keeping in currents, battery de-rating at the hovering depth, and the energy required to maintain a human cabin microclimate for crew comfort without dehumidifiers.
Harbor testing in Chennai, February 2025.
Testing
The Shakthi module was validated and refined during the experiment conducted in the Bay of Bengal in 2024.
Harbor wet tests were conducted in January and February 2025 at a shipbuilding port in Chennai. The focus was on the oxygen consumption pattern specific to the three-person crew during testing and the hydrodynamic performance of Matsya. The data were used to refine the Prana and Pushar models.
The AI models for the modules will continue to be refined based on the data from upcoming qualification phases, including the 500-m-depth demonstration planned in the first quarter of 2026 and subsequent activity in deeper waters.
After validation, Chaitanya will integrate a vehicle health management system that will assess subsystem health and the effect of the subsystems on each other and on Matsya as a whole. The system will be able to predict potential failures before they become critical.
The four CDT modules will make up the digital twin co-pilot. The goal of this work is to create a situational awareness system that incorporates machine learning and AI-driven insights to support crew during an underwater emergency.
Dr N. Vedachalam is a scientist and the project director of Matsya6000 at India’s National Institute of Ocean Technology.
Dr. VBN Jyothi is a scientist at India’s National Institute of Ocean Technology.
Dr. R. Balaji is the director of India’s National Institute of Ocean Technology.