ESA Tests Teleoperated Robots for Deep Space Communication Delays
Signal latency is forcing Europe’s top space scientists to rethink autonomy and robot control strategies for exploration
Humanoid robots may not be planting flags on Mars anytime soon, but they could quickly assist with building lunar infrastructure or performing delicate repairs in orbit.
The most pressing challenge isn't sending machines to space—it's controlling them when they're millions of kilometers away. The European Space Agency (ESA) is betting on a hybrid approach, combining real-time teleoperation with increasing levels of autonomy, to bridge this vast distance and enable effective human-robot collaboration.
One of the key issues lies in communication latency. For robots in low Earth orbit, latency can be as low as 20 milliseconds. However, those delays increase drastically when operations extend to the Moon or Mars. Operating a lunar robot from Earth introduces a lag of three to five seconds. On Mars, it balloons to a round-trip delay ranging from 8 to 40 minutes.
This isn't just a technical nuisance—it's a mission-critical limitation.
"The speed of light and processing times have a significant impact," said Thomas Krueger, Team Lead at ESA's Human Robot Interaction Lab. "If autonomy isn't mature enough, some operations from Earth become very painful. And for Mars, direct teleoperation is practically impossible."
So, the ESA is testing new paradigms: robots with limited autonomy operated from nearby spacecraft, human-machine interfaces fine-tuned for space, and software that lets astronauts supervise rather than micromanage.
Ground Lessons, Space Solutions
Krueger spoke during the Humanoids Summit in London on May 29, sharing ESA’s decade-long journey to develop space-capable teleoperation systems.
He began with a retrospective: In 2011, ESA had just started testing ways to control robots from orbit. Exoskeletons were the initial interface of choice, but these were soon swapped for multi-purpose haptic devices better suited for zero gravity.
A turning point came with a series of experiments aboard the International Space Station.
"We built a demonstrator and tested how we perceive moving into contacts, stiffnesses, from space and ground," Krueger said.
One important takeaway is that astronauts in microgravity lose some sensitivity. "We need to tune the force feedback gains if we want haptics to work well from space."
ESA's experimental setups evolved rapidly. One such mission involved an astronaut remotely operating a robot from orbit to complete tasks in a controlled Earth environment. While the experiments were limited—one had the astronaut controlling a ground robot from a hotel room due to logistical constraints—they offered valuable insights into both hardware and human-machine dynamics.
Supervising from the Moon
ESA's long-term vision embraces a tiered system: some robots will be directly controlled by astronauts aboard lunar-orbiting stations like the Lunar Gateway; others will run on semi-autonomous protocols; and some tasks will be executed fully autonomously.
The Lunar Gateway is a planned space station that will orbit the Moon as part of NASA's Artemis program. It's designed to be a multi-purpose outpost supporting lunar surface missions and scientific research in lunar orbit and serving as a launchpad for deeper space exploration.
"We want to try this scenario," Krueger explained, pointing to a robot built as a functional demonstrator, not for flight but to test concepts like locomotion, manipulation, and sensing in realistic mock environments.
An astronaut operates the robot using a haptic interface, and real-world testing took place in extreme terrains like Mount Etna in Italy to simulate extraterrestrial geology missions.
This layered approach could balance precision, safety, and efficiency. Astronauts could step in for delicate operations or emergencies, while autonomy takes over for routine or time-intensive tasks.
ESA also acknowledges that space-based robotics can't rely on a one-size-fits-all model. Different missions require different machines.
"I always imagine a construction site on Earth: we have so many different vehicles. The same will apply to a megastructure in orbit," Krueger said. "We need robots with a lot of different kinematic structures."
He also emphasized co-design: "The megastructure in orbit and the robots that construct it need to be co-designed. It's quite complex and challenging."
Not all space robotics will resemble humans—and that’s by design. While humanoids offer advantages in environments tailored for human use, their limitations become apparent in novel terrains or large-scale construction.
"If I want to build a canal, I can have 10 humanoids with shovels, but it wouldn't be smart," Krueger said. "I'd rather have an autonomous excavator."
Still, there are valid use cases for humanoid-like forms. Station keeping, for instance, where the environment is already built for human manipulation, makes sense.
"Everything that can be manipulated by a human would make sense. The robot can do the same. So the humanoid shape, or at least hands and arms, would be nice," he added.
ESA is currently collaborating with Germany’s space agency DLR to explore multi-robot control strategies. These include a dog-like robot named Bert for cave exploration, a heavy-duty unit for transportation, and humanoid systems for complex tasks. The goal is to develop systems that can switch between different levels of autonomy and human control depending on the situation.
Engineering Priorities and Market Realities
Krueger’s realistic outlook on the near-term future of space robotics reflects a careful balance between aspiration and engineering limitations.
"There are a lot of plans," he noted, referring to space-based solar power and orbital construction. "But as an engineer who worked in power systems, I'm a bit skeptical."
His skepticism isn't rooted in disbelief, but in the enormous technical and economic challenges. Launching hardware, maintaining communications, and building supporting infrastructure require massive investment.
"If autonomy is not there yet, we could indeed control them from the Lunar Gateway," he suggested.
But he quickly added that ESA remains flexible: "We don't have to use teleoperation. We can use it if it's required directly from the astronaut. We can use it from the ground if it's less painful. Or we can go autonomous where we can."
On the broader market outlook, Krueger sees the shift from public to private initiatives as a good sign.
"With robotics and space, this is now really similar to computers in the 1960s—from government to private projects," he said. "If companies understand they want and need to make money, and if it becomes financially interesting, then I'd be happy to see companies driving these aspects."
The Road to Multipurpose Robotics
ESA's future plans aren't limited to just proof-of-concept demos. Their work increasingly focuses on orchestrating multiple robots in collaborative roles, each optimized for specific functions.
"We know it's not only one robot," Krueger said. "We need to control multiple robots."
With DLR, ESA is planning experiments that involve dog-like scouts, cargo movers, and humanoids, testing how operators can manage them across different control schemes.
This could be the foundation for future missions that require everything from sample collection and base construction to infrastructure maintenance and deep exploration. Ultimately, the goal is seamless cooperation between humans and machines across varied distances, environments, and time lags.
As Krueger put it: "There are limits between direct operation, supervised autonomy, and full autonomy. Understanding where and how to apply each is the key to making robots truly useful in space."
In the next phase, expect more real-world testing, refined force-feedback systems, and broader industry participation. ESA isn’t building science fiction robots. It’s building the operational framework for a new era of exploration, where astronauts may only need to reach as far as the control panel to shape distant worlds.