Bio-Inspired Underwater Devices & Swarming Algorithms for Robotic Vehicles



Assistant Professor Wim van Rees and his team develop a simulation of a self-propelled undulating swimmer to better understand how deformable fins like fish can improve the propulsion of underwater appliances. Did. Credit: Image courtesy of MIT van Rees Lab

MIT Marine and mechanical engineers are taking advantage of advances in scientific computing to address and seize the opportunities of many ocean challenges.

Few environments are as merciless as the sea. Its unpredictable weather patterns and restricted communications left a large area of ​​the ocean unexplored and mysterious.

“The ocean is a fascinating environment, and there are many challenges today, including microplastics, algae outbreaks, coral bleaching, and rising temperatures,” said Wim van Rees, MIT’s professor of ABS career development. .. “At the same time, there are countless opportunities in the ocean, from aquaculture to energy harvesting to the exploration of many marine life that we have not yet discovered.”

Marine engineers and mechanical engineers like Van Reese are taking advantage of advances in scientific computing to tackle and seize many marine challenges. These researchers are developing technologies to better understand our oceans and how both biological and artificial vehicles can move in the ocean from microscale to macroscale. I understand.

Self-propelled undulating swimmer

Assistant Professor Wim van Rees and his team develop a simulation of a self-propelled undulating swimmer to better understand how deformable fins like fish can improve the propulsion of underwater appliances. Did. Credit: Image courtesy of MIT van Rees Lab

Bio-inspired underwater device

A complex dance is performed when the fish throws water. The flexible fins flap in the stream of water, leaving traces of eddy in the wake.

“Fish have complex internal muscle tissue to adapt to the exact shape of the body and fins, which goes far beyond what an artificial vehicle can do in terms of mobility, agility, and adaptability. It can be promoted in a variety of ways, ”explains van Rees.

According to van Rees, advances in layered modeling, optimization techniques, and machine learning bring us closer than ever to replicating flexible, morphing fish fins for use in underwater robotics. Therefore, it is necessary to understand how these soft fins affect propulsion.

Van Rees and his team are developing and using a numerical simulation approach to explore the design space for underwater devices with increased degrees of freedom for deformable fins, such as fish.

Prediction of loop current vortex

Graduate students Abhinav Gupta and Professor Pierre Lermusiaux have developed a new machine learning framework to make up for the lack of resolution and accuracy of existing dynamic system models. These frameworks can be used in many applications, such as improving the prediction of loop current eddies around the Gulf of Mexico oil rig. Credit: Image courtesy of MIT MSEAS Lab

These simulations help the team better understand the interaction between the hydrodynamic and structural mechanics of the soft and flexible fins of the fish as they move through the fluid flow. As a result, they can better understand how deformation of the fin shape impairs or improves swimming performance. “By developing accurate numerical methods and scalable parallel implementations, we can use supercomputers to solve exactly what is happening in this interface between flow and structure,” van Rees adds. ..

van Rees aims to develop a new generation of automated design tools for autonomous underwater devices by combining flexible underwater structure simulation algorithms with optimization and machine learning techniques. This tool helps engineers and designers develop, for example, robot fins and underwater vehicles. Robot fins and underwater vehicles can be smartly adapted to shape to better achieve immediate operational goals, such as swimming faster and more efficiently and performing maneuvering operations.

“Use this optimization and AI to reverse design across the parameter space to create smart, adaptable devices from scratch, or use accurate individual simulations to shape one shape to another. You can identify the physical principles that determine why it’s better than, “van Rees explains. ..

Robot vehicle swarming algorithm

Like Van Reese, Principal Investigator Michael Benjamin wants to improve the way vehicles move in the water. In 2006, Benjamin, then a postdoc at MIT, launched an open source software project for his autonomous helm technology. Used by companies such as Sea Machines, BAE / Riptide, Thales UK, Rolls Royce, and the US Navy, this software uses a new method of multi-objective optimization. Developed by Benjamin in his PhD, this optimization technique allows the vehicle to autonomously choose the direction, speed, depth and direction to go in order to reach multiple simultaneous goals.

Automated guided vehicle warming algorithm

Michael Benjamin has developed a swarming algorithm that allows unmanned vehicles like the one in the picture to be distributed optimally and avoid collisions. Credit: Michael Benjamin

Benjamin is now taking this technology one step further by developing algorithms for swarm and obstacle avoidance. These algorithms allow dozens of unmanned aircraft to communicate with each other and explore specific parts of the ocean.

First, Benjamin is looking at ways to best distribute self-driving cars into the ocean.

“Suppose you want to launch 50 vehicles in a part of the Sea of ​​Japan. What you want to know is, does it make sense to unload all 50 vehicles at once, or is it specific in a particular area? Does it make sense to have the mothership drop off at the point? ”Benjamin explains.

He and his team have developed an algorithm to answer this question. Using herd technology, each vehicle periodically communicates its location to other vehicles nearby. Benjamin’s software allows these vehicles to be optimally distributed in the parts of the ocean where they are operating.

At the heart of the success of swarming vehicles is the ability to avoid collisions. Collision avoidance is complicated by the International Maritime Regulations known as COLEGS or “Collision Regulations”. These rules determine which vehicle has the “right of way” when crossing the path, and poses a unique challenge to Benjamin’s swarming algorithm.

Although COLREGS was created from the perspective of avoiding another single contact, Benjamin’s herd algorithm had to take into account multiple unpiloted vehicles that would prevent them from colliding with each other.

To address this issue, Benjamin and his team have created a multi-object optimization algorithm that ranks specific operations on a scale from 0 to 100. 0 means a direct collision and 100 means the vehicle completely avoids the collision.

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

Researchers like Benjamin and van Rees use machine learning and multi-objective optimization to address the complexity of vehicles moving in marine environments, while researchers like Pierre Lermusiaux, Professor Nam Pyo Suh of MIT, Use machine learning to better understand the marine environment itself.

Improved ocean modeling and forecasting

The ocean is probably the best example of what is known as a complex dynamical system. Fluid dynamics, changing tides, meteorological patterns, and climate change make the ocean a different and unpredictable environment from moment to moment. The ever-changing nature of the marine environment can be very difficult to predict.

Researchers have used dynamic system models to predict the marine environment, but as Lermusiaux explains, these models have limitations.

“When developing a model, we cannot explain every molecule of ocean water with resolution. Accuracy The number of models and ocean measurements are limited. There may be model data points every 100 meters and every kilometer. Or, if you look at the world’s ocean climate model, there could be data points every 10 kilometers. This can have a significant impact on the accuracy of the forecast, “explains Lermusiaux.

Graduate students Abhinav Gupta and Lermusiaux have developed a new machine learning framework to make up for the lack of resolution and accuracy of these models. Their algorithm employs a simple model with low resolution, can fill the gap, and emulates a more accurate and complex model with high resolution.

The Gupta and Lermusiaux frameworks are the first to learn and implement the time delays of existing fitted models to improve predictive capabilities.

“Natural things don’t happen instantly, but all common models assume that things are happening in real time,” says Gupta. “To make the fitted model more accurate, the machine learning and data we enter into the equations must represent the impact of past states on future predictions.”

The team’s “nerve closure model” that explains these delays has the potential to improve predictions such as loop current vortices colliding with oil rigs in the Gulf of Mexico and the amount of phytoplankton in certain parts of the ocean. ..

As computing technologies such as Gupta and Lermusiaux’s neural closure models continue to advance, researchers can unravel the mysteries of the ocean and develop solutions to many of the challenges facing our ocean.

(function(d, s, id){
var js, fjs = d.getElementsByTagName(s)[0];
if (d.getElementById(id)) return;
js = d.createElement(s); = id;
js.src = “//”;
fjs.parentNode.insertBefore(js, fjs);
}(document, ‘script’, ‘facebook-jssdk’));

Bio-Inspired Underwater Devices & Swarming Algorithms for Robotic Vehicles Source link Bio-Inspired Underwater Devices & Swarming Algorithms for Robotic Vehicles

The post Bio-Inspired Underwater Devices & Swarming Algorithms for Robotic Vehicles appeared first on California News Times.

Source link


Please enter your comment!
Please enter your name here

Share post:


More like this

FOIA Reveals Postal Inspectors Monitored Conservative Protestors

According to records obtained by The Washington Times, the...

Solana posts gains of 10% in a day, here is why the move can continue higher

Disclaimer: The findings of the following analysis are the sole...

“They’re Panicked Because They Know the Current System is Doomed”

"They're Panicked Because They Know the Current System is...