The team completed a sample collection task by controlling a virtual simulation, which then sent commands to a physical robot to mirror the simulation’s actions. They did so while monitoring the simulation without camera streams, making the tool particularly useful for delayed teleoperation on the Moon.
Several public and private organisations are researching how best to extract resources, such as oxygen and water, from readily available materials such as lunar regolith (moon dust).
Remote handling of regolith will be an essential step in these activities, as it would first need to be collected from the Moon’s surface. However, regolith is not easy to work with because it is sticky, abrasive, and will be handled under reduced gravity.
In a statement, lead author Joe Louca from Bristol’s School of Engineering Mathematics and Technology, and the Bristol Robotics Laboratory, said: “One option could be to have astronauts use this simulation to prepare for upcoming lunar exploration missions.
“We can adjust how strong gravity is in this model, and provide haptic feedback, so we could give astronauts a sense of how Moon dust would feel and behave in lunar conditions – which has a sixth of the gravitational pull of the Earth’s.
“This simulation could also help us to operate lunar robots remotely from Earth, avoiding the problem of signal delays.”
Using a virtual model of regolith can also reduce the barriers to entry for people looking to develop lunar robots. Instead of needing to invest in expensive simulants, or have access to facilities, people developing lunar robots could use this simulation to carry out initial tests on their systems.
Now, the team will investigate how people respond to this system when controlling a robot with several seconds of delay. Systems with human operators that are technically effective may still have to overcome non-technical barriers, like whether a person trusts the system will work.
Louca said: “The model predicted the outcome of a regolith simulant scooping task with sufficient accuracy to be considered effective and trustworthy 100 per cent and 92.5 per cent of the time.
“In the next decade we’re going to see several crewed and uncrewed missions to the Moon, such as NASA’s Artemis program and China’s Chang’e program.
“This simulation could be a valuable tool to support preparation or operation for these missions.”
The testing was carried out at the European Space Agency’s European Centre for Space Applications and Telecommunications site in Harwell, Oxfordshire.
The team’s paper - ‘Demonstrating Trustworthiness in Open-Loop Model Mediated Teleoperation for Collecting Lunar Regolith Simulant’ – was presented at the IROS 2024, the EEE/RSJ International Conference on Intelligent Robots and Systems.
Comment: Will AI take all the engineering jobs or improve productivity
The human oversight will be a problem. For this to be effective the human requires extensive experience and knowledge in addition to the creative...