The ocean is so vast and vast. There are limits to human efforts and abilities. We can also only explore the ocean to a certain depth. Recently, scientists released a robot from the lab at Caltech that can go deeper and perform more relevant underwater tasks.
CARL-Bot
“We are envisioning an approach to global ocean exploration like sending a swarm of robots out into the ocean to explore the oceans,” said John O. Dabiri, a professor of aeronautics and mechanical engineering at the California Institute of Technology. track, aimed at climate change, to uncover the mysteries of the ocean.”
CARL-Bot was born with the mission to explore the ocean. Photo: Caltech.
And from there CARL-Bot was born. It is a palm-sized aquatic robot that is motorized and has sensors that can detect pressure, depth, acceleration, and orientation. Everything CARL does is powered by a tiny 1 megabyte microcontroller inside.
CARL is the latest transatlantic innovation in Dabiri’s lab created by Caltech graduate student Peter Gunnarson . The first tests Gunnarson ran with it were in his bathtub, since Caltech labs closed in early 2021 because of COVID.
This robot is learning how to survive in the ocean
Currently CARL can still be controlled remotely. But when CARL actually reaches the deepest parts of the ocean, there’s no way to control it. That means no researcher can give CARL directions — it needs to learn to navigate the mighty ocean on its own. Gunnarson and Dabiri reached out to computer scientist Petros Koumoutsakos, who helped develop AI algorithms for CARL, teaching it to orient itself based on changes in the immediate environment and from past experiences.
CARL is programmed to improvise situations in the ocean. Photo: Azorobotics.
The set of algorithms developed by Koumoutsakos can perform pathfinding calculations on the small robot. Algorithms take advantage of the robot’s memory of the journey, like how to navigate a vortex or ocean current. “We can use that information to decide how to navigate those situations in the future,” explains Dabiri.
CARL’s program allows it to remember the same paths it took on previous missions. “Through repeated experience, it got better and better at sampling the ocean with less time and less energy”.