The next generation of Mars and Moon exploration may look a little more like a household pet than a traditional rover, with robotic “dogs” one day walking across alien terrain instead of rolling over it.
In a new study published in Frontiers in Space Technologies, researchers have demonstrated that a four-legged robotic system equipped with advanced scientific instruments can autonomously explore, analyze, and identify geological targets in environments designed to mimic the surfaces of Mars and the Moon.
The findings underscore a fundamental transformation in planetary exploration: moving from slow, human-directed missions to agile, semi-autonomous technologies that can efficiently traverse and study the toughest terrains in the solar system.
“Future planetary exploration missions require advanced robotic capabilities to efficiently identify and characterize lithologies, rock textures, and mineralogies for astrobiological investigations and in-situ resource utilization (ISRU),” researchers write. “This study explores an alternative semi-autonomous, multi-target exploration strategy using a legged robotic system, which could enable faster, more efficient surface prospecting on the Moon and Mars.”
In the study, researchers used “ANYmal,” a four-legged robot built by Anybotics and specially equipped with a robotic arm carrying a microscopic imager and a Raman spectrometer—tools capable of examining rock textures and identifying mineral composition on site.
Unlike traditional wheeled rovers, which struggle on steep or unstable terrain, legged robots, like ANYmal, can access regions that have long remained out of reach, including crater walls, fractured landscapes, and ice-rich slopes.
Researchers argue that the capabilities demonstrated in this study are important for advancing efficient, independent exploration, especially in environments where communication with Earth is limited.
A Bottleneck in Space Exploration
For decades, robotic exploration has been limited not just by engineering challenges but also by physics.
On Mars, communication delays between Earth and robotic missions can range from 3 to 22 minutes one way. That means every movement, measurement, or decision must be carefully planned in advance, often resulting in painfully slow progress. Even the most advanced rovers typically travel only a few dozen to a few hundred yards per day while scientists analyze incoming data and prepare the next set of commands.
This limitation becomes even more severe for missions seeking to explore regions farther from the solar system. Future missions to places like Saturn’s moon Titan, where delays exceed an hour, will require robots capable of making decisions largely on their own.
Researchers attempt to address this problem by testing whether a robot can autonomously identify and analyze multiple scientific targets within a single mission cycle, rather than relying on step-by-step human supervision.
Testing Mars and Moon with anymal on Earth
To replicate real-world conditions, the team conducted a series of analog missions within controlled environments that mimicked Martian and lunar terrain.
In a laboratory known as the “Marslabor” at the University of Basel, researchers recreated Mars surface conditions using basalt fragments and hematite-rich materials. Lighting conditions were carefully adjusted to replicate the balance of direct and diffuse sunlight found on Mars.
For the lunar simulations, the team conducted experiments in near darkness to mimic the extreme lighting conditions near the Moon’s southern polar regions, where extended shadows and high contrast can obscure surface features.
The robot was tasked with identifying and analyzing various rock types known to exist on Mars and the Moon, including gypsum, carbonate rocks, basalt, dunite, and anorthosite—materials that are scientifically significant for analyzing planetary history and assessing potential resources.
Two Ways to Explore Another World
The study compared two distinct exploratory strategies. In the first, a semi-autonomous, multi-target approach, human operators selected multiple targets at the start of the mission. The ANYmal robot then carried out all navigation, positioning, and scientific measurements independently.
In the second, operators used the more traditional method of guiding the ANYmal robot step by step, reviewing data after each measurement and deciding what to do next.
The results showed that across four Mars-like missions, the semi-autonomous system successfully identified targets at rates of 66.7%-100%, with mission durations as short as 12 minutes.
By contrast, the human-guided lunar mission achieved a perfect 100% success rate—but took significantly longer, requiring 41 minutes to complete. The trade-off highlights a key tension in future exploration: speed versus precision.
Seeing and “Touching” Alien Worlds
A major innovation in the system is the combination of two complementary instruments.
A custom-built microscopic imager captures high-resolution close-up images of rock textures, revealing features like grain size, layering, and structural patterns. Meanwhile, the Raman spectrometer mounted on the robot’s arm uses laser-based spectroscopy to identify the mineral composition of samples by detecting characteristic spectral peaks (distinctive markers in the spectrum produced when the laser light interacts with mineral molecules).
Together, these tools allow the ANYmal robot to observe, hypothesize, and confirm findings, something remarkably close to what a human geologist would do in the field.
In the Martian simulations, the system successfully identified gypsum through a distinctive Raman peak at around 1010 cm⁻¹. Meanwhile, sulfur deposits were detected at approximately 473 cm⁻¹.
In lunar simulations, the robot identified olivine-rich dunite and titanium-bearing rutile, both of which are considered important for analyzing planetary formation and potential resource extraction.
Why ANYmal’s Legs Matter
Perhaps the most significant implication of the study lies not in the instruments, but in the ANYmal robot itself.
Traditional rovers, like NASA’s Curiosity and Perseverance, rely on wheels—an effective but limited solution. Steep slopes, loose regolith, and rocky obstacles can quickly halt their progress.
By contrast, legged robots, like ANYmal, can step over obstacles, maintain balance on uneven terrain, and access areas that would be inaccessible to wheeled systems.
This could open up entirely new regions for exploration, including sites most likely to preserve signs of past life—such as sedimentary layers, hydrothermal systems, and ice-rich environments.
A Shift Toward Autonomous Science
The study envisions robots evolving from remote-controlled tools into autonomous scientific agents.
By allowing machines to select, analyze, and prioritize targets with minimal human intervention, missions could cover more ground, gather more data, and respond dynamically to new discoveries—all while operating millions of miles from Earth.
Researchers emphasize that this approach will be essential as missions push deeper into the solar system, where communication delays make real-time control impossible.
Their results suggest that semi-autonomous systems could dramatically increase the efficiency of planetary exploration, without sacrificing scientific value.
While the results are promising, the study also highlights challenges.
Environmental factors such as dust, lighting conditions, and instrument positioning can affect data quality. In some cases, blurred images or noisy spectral readings limited the system’s ability to accurately identify targets.
Still, the overall success of the experiments suggests that these obstacles are achievable and that the benefits of semi-autonomous exploration may far outweigh the risks.
Ultimately, as space agencies prepare for a new wave of missions to the Moon, Mars, and beyond, autonomous four-legged robotic explorers like ANYmal could soon move from the realm of science fiction to the reality of planetary exploration.
“Our study demonstrates that a multi-target semi-autonomous exploration approach is a viable option for geological investigations in planetary surface missions where the inability to control a robot in real-time significantly slows down exploration times and, consequently, the scientific return of the mission,” researchers conclude. “These findings emphasize the need to balance mission automation, efficiency, and scientific return based on operational constraints and planetary environments.”
Tim McMillan is a retired law enforcement executive, investigative reporter and co-founder of The Debrief. His writing typically focuses on defense, national security, the Intelligence Community and topics related to psychology. You can follow Tim on Twitter: @LtTimMcMillan. Tim can be reached by email: tim@thedebrief.org or through encrypted email: LtTimMcMillan@protonmail.com