Bio-inspired underwater devices and swarm algorithms for robotic vehicles


Assistant professor Wim van Rees and his team have developed simulations of undulating self-propelled swimmers to better understand how fish-like deformable fins could improve propulsion in underwater devices, seen here in a top view. Image credit: Image courtesy of the MIT van Rees Lab

WITH Marine and mechanical engineers use advances in scientific computing to address the ocean’s many challenges and seize its opportunities.

There are few environments as unforgiving as the ocean. Its unpredictable weather patterns and limitations in communications have left large swathes of the ocean unexplored and shrouded in mystery.

“The ocean is a fascinating environment with a number of current challenges such as microplastics, algal blooms, coral bleaching and rising temperatures,” says Wim van Rees, ABS Career Development Professor at MIT. “At the same time, the ocean holds countless opportunities – from aquaculture to energy generation to studying the many marine life that we have not yet discovered.”

Marine engineers and mechanical engineers like van Rees use advances in scientific computing to address the ocean’s many challenges and seize its opportunities. These researchers are developing technologies to better understand our oceans and how both organisms and man-made vehicles can move within them, from the micro to the macro scale.

Self-propelled wave swimmers

Assistant professor Wim van Rees and his team have developed simulations of self-propelled wave swimmers to better understand how fish-like deformable fins could improve propulsion in underwater devices, seen here as two fish lying side by side. Image credit: Image courtesy of the MIT van Rees Lab

Bio-inspired underwater devices

An intricate dance takes place as fish dart through the water. Flexible fins flap in water currents, leaving a trail of eddies.

“Fish have intricate internal musculature to accommodate the precise shape of their bodies and fins. This allows them to move in many different ways, well beyond what an artificial vehicle can do in terms of manoeuvrability, agility or adaptability,” explains van Rees.

According to van Rees, thanks to advances in additive manufacturing, optimization techniques and machine learning, we are closer than ever to replicating flexible and changing fish fins for use in underwater robotics. Therefore, there is a greater need to understand how these soft fins affect propulsion.

Van Rees and his team develop and use numerical simulation approaches to explore the design space for underwater devices that exhibit an increase in degrees of freedom, such as fish-like deformable fins.

Loop current eddy prediction

PhD student Abhinav Gupta and Professor Pierre Lermusiaux have developed a new machine learning framework to compensate for the lack of resolution or accuracy in existing dynamic system models. Your frame can be used for a number of applications, including improved predictions of loop current vortices around oil platforms in the Gulf of Mexico. Image credit: Image courtesy of the MIT MSEAS Lab

These simulations help the team better understand the interplay between the fluid and structural mechanics of the soft, flexible fish fins as they move through a fluid flow. This allows them to better understand how fin shape deformations can affect or improve swimming performance. “By developing accurate numerical techniques and scalable parallel implementations, we can use supercomputers to resolve what exactly is happening at this interface between flow and structure,” adds van Rees.

By combining his simulation algorithms for flexible underwater structures with optimization and machine learning techniques, van Rees aims to develop an automated design tool for a new generation of autonomous underwater devices. This tool could help engineers and designers create things like robotic fins and underwater vehicles that can intelligently adjust their shape to better meet their immediate operational goals — be it swimming faster and more efficiently, or manoeuvring.

“We can use this optimization and AI to perform inverse design within the full parameter space and build intelligent, adaptive devices from scratch, or use accurate custom simulations to identify the physical principles that determine why a shape is better.” performs better than another,” explains van Rees.

Swarm algorithms for robotic vehicles

Like van Rees, lead research scientist Michael Benjamin wants to improve how vehicles maneuver through the water. In 2006, then a postdoc at MIT, Benjamin started an open source software project for an autonomous control technology he developed. The software, which has been used by companies such as Sea Machines, BAE/Riptide, Thales UK and Rolls Royce, and the United States Navy, uses a novel method of multi-objective optimization. This optimization method, developed by Benjamin during his PhD, allows a vehicle to autonomously choose the course, speed, depth and direction to travel in order to achieve multiple simultaneous destinations.

Swarm algorithms for unmanned vehicles

Michael Benjamin has developed swarm algorithms that allow unmanned vehicles like the one shown to disperse optimally and avoid collisions. Credit: Michael Benjamin

Now Benjamin is taking this technology a step further by developing swarm and obstacle avoidance algorithms. These algorithms would allow dozens of unmanned vehicles to communicate with each other and explore a specific part of the ocean.

First, Benjamin investigates how autonomous vehicles can best be distributed in the ocean.

“Let’s say you want to launch 50 vehicles in a stretch of the Sea of ​​Japan. We want to know: does it make sense to drop all 50 vehicles in one place or have a mother ship drop them off at specific points in a specific area? explains Benjamin.

He and his team have developed algorithms that answer this question. Using swarm technology, each vehicle periodically shares its location with other nearby vehicles. Benjamin’s software allows these vessels to disperse in an optimal distribution for the part of the ocean in which they operate.

Central to the success of the swarming vehicles is the ability to avoid collisions. Collision avoidance is complicated by international maritime regulations known as COLREGS – or “Collision Regulations”. These rules determine which vehicles have “priority” when crossing paths, presenting a unique challenge to Benjamin’s swarm algorithms.

The COLREGS are written from the perspective of avoiding another single contact, but Benjamin’s swarm algorithm had to account for multiple unmanned vehicles trying to avoid colliding with each other.

To address this problem, Benjamin and his team developed a multi-object optimization algorithm that scores specific maneuvers on a scale from zero to 100. A zero would be a direct collision, while 100 means the vehicles avoid a collision entirely.

“Our software is the only marine software where multi-objective optimization is the central mathematical basis for decision-making,” says Benjamin.

While researchers like Benjamin and van Rees are using machine learning and multi-objective optimization to address the complexities of vehicles moving through marine environments, others like Pierre Lermusiaux, MIT’s Nam Pyo Suh professor, are using machine learning to better understand the marine environment itself .

Improvement in ocean modeling and forecasting

Oceans are perhaps the best example of a so-called complex dynamical system. Fluid dynamics, changing tides, weather patterns, and climate change make the ocean an unpredictable environment that changes from one moment to the next. The ever-changing nature of the marine environment can make predictions incredibly difficult.

Researchers have used dynamic systems models to make predictions for ocean environments, but as Lermusiaux explains, these models have limitations.

“You cannot consider every water molecule in the ocean when developing models. The resolution and accuracy of models and the ocean measurements are limited. There could be a model data point every 100 meters, every kilometer, or if you look at global ocean climate models, you could have a data point about every 10 kilometers. This can have a major impact on the accuracy of your prediction,” explains Lermusiaux.

PhD student Abhinav Gupta and Lermusiaux have developed a new machine learning framework to compensate for the lack of resolution or accuracy in these models. Their algorithm takes a simple, low-resolution model and can fill in the gaps by emulating a more accurate, more complex, high-resolution model.

For the first time, Gupta and Lermusiaux’s framework learns and introduces time delays into existing approximation models to improve their predictive abilities.

“Things in nature don’t happen instantly; However, all current models assume that things happen in real time,” says Gupta. “To make an approximate model more accurate, the machine learning and the data you feed into the equation must represent the impact of past states on the future prediction.”

The team’s “neural closure model,” which accounts for these delays, could potentially lead to improved predictions for things like a loop current vortex hitting an oil rig in the Gulf of Mexico, or the amount of phytoplankton in a given part of the ocean.

As computational technologies like Gupta and Lermusiaux’s neural closure model continue to improve and evolve, researchers can begin to unravel more of the ocean’s mysteries and develop solutions to the many challenges our oceans face.

Comments are closed.