AI uses reinforcement learning to navigate the oceans.
Feb 14, 2024 1:20:52 GMT -5
Post by sabbirislam258 on Feb 14, 2024 1:20:52 GMT -5
Engineers at Caltech, ETH Zurich, and Harvard are working on an artificial intelligence (AI) that could enable autonomous drones to use ocean currents to aid in their navigation. With this approach, the drone does not have to fight the current. The research was published in Nature Communications on December 8. John O'Dubry is the Centennial Professor of Aeronautics and Mechanical Engineering and one of the authors of the study. "When we want robots to explore the deep ocean, especially in sheep, it's almost impossible to control them with a joystick from 20,000 feet on the surface. We don't even give them data about local ocean currents.
They need to be able to navigate because we can't detect them New Zealand Telemarketing Data from the surface. Instead, at a certain point we need the ocean-born drones to decide for themselves how to move forward. Yes," says Daberi. Testing the AI Engineers tested the AI's accuracy with computer simulations, and the team developed a small robot that runs the algorithm on a computer chip, which could power maritime drones on Earth as well as other planets. Eventually, they could develop an autonomous system that monitors the state of the planet's oceans, and it would do so in combination with artificial equipment previously developed to help jellyfish swim on command.
For this approach to work, drones must decide on their own where to go and how to get there. They will likely have to rely on data they collect themselves, which will be in the form of information about the water currents they are experiencing. The researchers used a reinforcement learning network to solve this, and they wrote software that could run on a tiny microcontroller. The team was able to use computer simulations to teach the AI to navigate. The simulated swimmer only had access to information about water currents at his immediate location, but he was able to quickly learn how to take advantage of eddies in the water to navigate toward a target.
They need to be able to navigate because we can't detect them New Zealand Telemarketing Data from the surface. Instead, at a certain point we need the ocean-born drones to decide for themselves how to move forward. Yes," says Daberi. Testing the AI Engineers tested the AI's accuracy with computer simulations, and the team developed a small robot that runs the algorithm on a computer chip, which could power maritime drones on Earth as well as other planets. Eventually, they could develop an autonomous system that monitors the state of the planet's oceans, and it would do so in combination with artificial equipment previously developed to help jellyfish swim on command.
For this approach to work, drones must decide on their own where to go and how to get there. They will likely have to rely on data they collect themselves, which will be in the form of information about the water currents they are experiencing. The researchers used a reinforcement learning network to solve this, and they wrote software that could run on a tiny microcontroller. The team was able to use computer simulations to teach the AI to navigate. The simulated swimmer only had access to information about water currents at his immediate location, but he was able to quickly learn how to take advantage of eddies in the water to navigate toward a target.