NASA’s Jet Propulsion Laboratory is testing a versatile robot that will autonomously map, navigate and explore previously inaccessible targets. How to create a robot that can independently visit places that no one has seen before, in real time without human intervention? A team building a snake-like robot to navigate rough terrain at NASA’s Jet Propulsion Lab is tackling this challenge with a beginner’s mindset: build fast, test often, learn, tune, iterate.

Dubbed EELS (short for Exobiology Extant Life Surveyor), the self-propelled autonomous robot was inspired by its desire to descend into narrow openings on the surface to search for signs of life in the ocean hiding under the icy crust of Saturn’s moon Enceladus. throwing geysers into space. Although testing and development continues, designing for such a complex purpose has resulted in a highly adaptable robot. EELS can safely navigate a wide variety of terrains on Earth, the Moon and beyond, including undulating sand and ice, rock walls, very steep craters for navigators, underground lava tubes and labyrinths inside glaciers.

Team members from JPL are testing a robot snake called EELS at a ski resort in the mountains of Southern California in February. Designed to detect the environment, calculate risks, travel and collect data in real time without human intervention, EELS can finally explore destinations throughout the Solar System. Credit: NASA/JPL-Caltech

“It has the ability to go where other robots can’t. While some robots are better on this or that type of terrain, the idea behind EELS is to do anything,” said Matthew Robinson, EELS project manager at JPL. You want to send a risk-aware robot that’s ready and able to make decisions on its own.”

The project team started building the first prototype in 2019 and has been making constant changes. Since last year, they have been conducting monthly field tests and developing both hardware and software that allow EELS to operate autonomously. In its current form, called EELS 1.0, the robot weighs about 220 pounds (100 kilograms) and is 13 feet (4 meters) long. It consists of 10 identical segments that rotate using screw threads for movement, traction and connection. The team tested a variety of screws: 3D-printed 8-inch (20 centimeters diameter) white plastic screws for testing in loose terrain, and narrower, sharper black metal screws for ice.

How does EELS think and act?

Due to communication latency between Earth and deep space, EELS is designed to autonomously sense the environment, calculate risks, travel, and collect data with as yet unspecified scientific instruments. When something goes wrong, the goal is for the robot to heal on its own without human assistance.

“Imagine a self-driving car, but there are no brake lights, no traffic lights, not even roads. The robot needs to understand what the road is and try to follow it,” said Rohan Thakker, head of the autonomous project. “Then it should come down from a 100-foot drop and not fall.”

EELS creates a 3D map of its surroundings using four pairs of stereo cameras and lidar, which is similar to radar but uses short laser pulses instead of radio waves. Navigation algorithms determine the safest route using data from these sensors. The goal was to create a library of “moves”, or ways for the robot to move in response to terrain challenges, from turning sideways to twisting on itself, a motion the team referred to as a “banana.”

In its final form, the robot will contain 48 actuators (essentially small motors), giving it the flexibility to adopt a variety of configurations, but adding complexity to both hardware and software teams. Tucker compares drivers to “48 steering wheels”. Many have a built-in force and torque sensor that works like a kind of leather, so the EELS can sense how much force is applied to the terrain. This helps it adjust to move vertically in narrow, rough-surfaced gutters and simultaneously push up opposing walls like a rock climber.