An AI-trained four-legged robot has learned the same lesson as the Apollo astronauts – that jumping could be the best way to move on the moon. An update on LEAP (Legged Exploration of the Aristarchus Plateau), an important concept study funded by the European Space Agency to explore some of the most challenging lunar terrain, was presented today at the Europlanet Science Conference (EPSC) 2022 in Granada by Patrick Bambach.
“The LEAP target is the Aristarchus Plateau, a region of the Moon that is particularly rich in geological features but extremely difficult to access,” said Patrick Bambach of the Max Planck Institute for Solar System Research in Germany. “Using the robot, we can investigate key features to study the geological history and evolution of the Moon, such as ejecta around craters, new impact sites, and collapsed lava tubes, where materials may not have changed due to space weathering and other processes.”
The LEAP team is working to integrate the robot into the European Space Agency’s Large European Logistics Landing Module (EL3), which is scheduled to land on the Moon several times from late 2020 to early 2030. LEAP is based on the two-legged robot, ANYmal, which was developed In ETH Zürich and the ANYbotics spin-off. It is currently being adapted to the lunar environment by a consortium of ETH Zurich, Max Planck Institute for Solar System Research, OHB, University of Münster and The Open University.
“Conventional rovers have enabled great discoveries on the Moon and Mars, but they have limitations,” Bambach said. “Exploring terrain with loose soil, large boulders, or slopes greater than 15 degrees is especially challenging using wheels. For example, its mission to Mars rover, Spirit, was terminated when it got stuck in the sand.”
The ANYmal can move in different walking gaits, enabling it to cover large distances in a short amount of time, climb steep slopes, deploy scientific instruments, and even recover in the unlikely event of a fall. The robot can also use its legs to dig channels into the soil, and overturn rocks or smaller boulders for further examination and sample capture.
Initially, the robot was trained using a reinforcement learning approach in a virtual environment to simulate the characteristics of lunar terrain, gravity, and dust. He was also deployed in the field for a walk in the fresh air.
“Interestingly, Animal began to use a kinematic position similar to jumping, just as the Apollo astronauts did — realizing that jumping can be more energy efficient than walking,” Bambach said.
The current design is still less than 100 kg and includes 10 kg of scientific payload mass, and could theoretically carry multispectral sensors, ground-penetrating radar, mass spectrometers, gravimeters, and other instruments.
“LEAP’s ability to collect select samples and bring them to a lander or ascent vehicle provides additional exciting opportunities for sample return missions in very challenging environments on the Moon or Mars,” Bambach said.
Disclaimer: AAAS and EurekAlert! is not responsible for the accuracy of newsletters sent on EurekAlert! Through the contributing institutions or for the use of any information through the EurekAlert system.