Legged Robots Pave the Way for Next-Gen Deep Space Exploration
As humanity’s ambitions in space grow bolder, so too does the need for more advanced robotic systems capable of navigating the most extreme and uncharted terrains beyond Earth. While wheeled rovers like NASA’s Curiosity and China’s Yutu have proven their worth on relatively flat lunar and Martian surfaces, future deep space missions are targeting regions far more complex—permanently shadowed craters at the Moon’s poles, steep cliffs on Mars, or the rubble-strewn surfaces of asteroids. These environments demand a new class of robotic explorer: the legged robot.
A comprehensive review published in China Mechanical Engineering analyzes the current state of legged robotics for deep space landing exploration, offering a detailed comparison of leading international prototypes and identifying the critical technological hurdles that must be overcome before these machines can become operational assets in real missions. The study, led by Sun Junkai from Jilin University in collaboration with engineers from the Beijing Institute of Spacecraft System Engineering and the Beijing Key Laboratory of Intelligent Space Robotic System Technology and Applications, provides a roadmap for the future of extraterrestrial mobility.
For decades, wheeled rovers have dominated planetary exploration. Their mechanical simplicity, energy efficiency, and stability on smooth terrain made them the natural choice for early missions. However, their limitations are becoming increasingly apparent. A rover can only go where its wheels can roll, and rocky outcrops, deep crevices, loose regolith, and steep inclines often present insurmountable barriers. This restricts scientific access to some of the most geologically and astrobiologically significant areas, such as the Moon’s south pole, where water ice is believed to be trapped in eternal darkness.
Legged robots, by contrast, offer a fundamentally different approach to mobility. Inspired by biological systems, they can step over obstacles, climb slopes, and adjust their gait in real time to maintain balance on uneven ground. This adaptability could dramatically expand the operational range of future missions, enabling access to previously unreachable scientific targets. The research by Sun and his colleagues highlights that while the concept is promising, the path to engineering a reliable, space-qualified legged robot is fraught with challenges.
The United States has been at the forefront of this research for over three decades. One of the earliest and most ambitious projects was the Ambler robot, developed in the late 1980s by Carnegie Mellon University and NASA’s Jet Propulsion Laboratory (JPL). Standing over six meters tall and weighing more than two metric tons, Ambler was a colossal machine designed for static walking on Mars-like terrain. Its legs operated on a Cartesian coordinate system, allowing it to lift its body vertically to clear large obstacles. Although it never flew, Ambler laid the groundwork for advanced terrain perception and gait planning, using laser scanners and force sensors to build 3D maps of its environment. However, its sheer size and weight made it impractical for actual spaceflight, and its control systems, while innovative for the time, lacked the autonomy needed for deep space operations where communication delays can stretch to minutes or even hours.
Building on this legacy, JPL developed the Dante series of robots in the 1990s. Dante II, an eight-legged machine, was tested in the harsh environment of Mount Spurr, an active volcano in Alaska. Weighing 770 kilograms, it was designed to explore volcanic craters on other planets. A key feature was its 300-meter tether, which provided power, data transmission, and physical stability on steep slopes. This allowed it to descend into rugged terrain while remaining connected to a base station. While the tether enabled remote operation over long distances, it also severely limited the robot’s range and maneuverability, effectively anchoring it to a single line of movement. The experiment demonstrated the potential of legged systems in extreme environments but also underscored the drawbacks of relying on physical connections in space, where untethered operation is essential.
The most advanced American concept to date is the All-Terrain Hex-Legged Extra-Terrestrial Explorer, or ATHLETE. First prototyped in 2005, ATHLETE represents a shift toward more practical, multi-functional systems. The initial SDM prototype weighed 800 kilograms and featured six legs, each with six degrees of freedom and a wheel at the end, enabling both walking and rolling. This dual-mode mobility allows it to move efficiently on flat ground and switch to a walking gait when encountering obstacles. Its successor, the T12, doubled in size and complexity, with seven degrees of freedom per leg and a payload capacity of 500 kilograms. ATHLETE is designed not just to move, but to perform complex tasks such as deploying habitats, transporting cargo, and assembling infrastructure—critical functions for future lunar bases.
What sets ATHLETE apart is its integration of multiple subsystems: high-resolution stereo cameras, inertial measurement units, and sophisticated software for autonomous navigation. It can generate a 3D virtual model of its surroundings and plan its path accordingly, reducing the need for constant human oversight. However, even ATHLETE faces significant hurdles. Its weight, complexity, and power consumption—up to 4000 watts—are substantial barriers to launch and operation. Moreover, its mechanical intricacy increases the risk of failure, a critical concern when repair is impossible.
Europe has taken a different approach, focusing on smaller, more agile, and often collaborative systems. The German Aerospace Center (DLR) and the German Research Center for Artificial Intelligence (DFKI) have been instrumental in this effort. One notable example is the SpaceClimber, a 23-kilogram hexapod robot designed to scale the walls of lunar craters. With 26 degrees of freedom and a bio-inspired control system that mimics neural oscillators, SpaceClimber can adapt its gait in real time to maintain stability. Its control architecture includes a central pattern generator for rhythmic motion and reflex-based responses to disturbances, allowing it to react quickly to unexpected terrain changes.
Even more innovative is the Rimres system, which combines a six-legged robot called Crex with a four-wheeled rover named Sherpa. These two robots can operate independently or dock together, sharing power, data, and even mechanical support. Sherpa can carry Crex to a difficult area and then deploy it for detailed inspection. This modular, reconfigurable design enhances mission flexibility and redundancy. If one robot fails, the other may still complete critical tasks. The system uses a standardized electromechanical interface (EMI) that allows for the attachment of various payloads, such as scientific instruments or additional batteries, further increasing its adaptability.
Another European project, Charlie, developed by DFKI, takes inspiration from primates. Weighing just 21.5 kilograms, Charlie can switch between quadrupedal and bipedal locomotion, allowing it to crawl through tight spaces or stand upright to manipulate objects. Its 36 degrees of freedom provide exceptional dexterity, but at the cost of increased mechanical complexity and reduced payload capacity. The robot uses a layered control system that includes a local feedback loop to compensate for disturbances, enabling stable movement even on shifting terrain.
Switzerland’s contribution comes in the form of SpaceBok, a 20-kilogram quadruped developed by ETH Zurich. Unlike walking robots, SpaceBok is designed for dynamic jumping, a highly efficient mode of locomotion in low-gravity environments like the Moon or asteroids. Using high-torque motors and spring-loaded legs, it can store energy during landing and reuse it for the next jump, minimizing power consumption. In simulated lunar gravity, it has achieved jumps of up to 1.3 meters. Its control system relies on virtual model control, allowing it to maintain balance during flight and landing phases. While still in the experimental phase, SpaceBok demonstrates the potential of non-traditional locomotion strategies for extraterrestrial exploration.
China, while entering the field more recently, has made rapid progress. The NOROS robot, developed at Beihang University, features a modular leg design that can switch between wheeled and legged modes. This hybrid approach aims to combine the efficiency of wheels with the adaptability of legs. Its passive joints and force sensors allow it to detect ground contact and adjust its posture, even recovering from a fall. However, the complexity of its reconfigurable legs increases the likelihood of mechanical failure.
Another Chinese prototype, the E1 Spider, is a heavy-duty hexapod designed for high load capacity. Weighing around 320 kilograms, it can carry payloads of up to 150 kilograms, making it suitable for transporting equipment or samples. Its robust leg structure and three active joints per leg provide strong terrain adaptability. The control system separates high-level planning from low-level joint control, a common architecture in advanced robotics. However, like many current systems, its autonomy is limited, relying heavily on human operators for decision-making.
Perhaps the most ambitious Chinese concept is the wheeled-leg mobile manned lunar lander. This vehicle integrates landing, ascent, and surface mobility into a single system. Its six legs can extend or retract, allowing it to transition between a low, stable configuration for landing and a high, mobile configuration for traversing rough terrain. In legged mode, it can climb obstacles up to four meters high—a capability unmatched by any existing rover. This design reflects a growing trend toward multi-functional spacecraft that can perform a variety of tasks without requiring separate landers and rovers.
Despite these impressive prototypes, none have yet been deployed on an actual mission. As Sun Junkai and his co-authors point out, the primary obstacles are not just technical but systemic. Current legged robots suffer from a lack of intelligent software that can operate autonomously under the constraints of deep space communication. The time delay between Earth and Mars can exceed 20 minutes, making real-time remote control impossible. Robots must therefore make decisions on their own, from path planning to obstacle avoidance to scientific target selection. While progress has been made in perception and control, true autonomy—where a robot can assess its environment, prioritize tasks, and execute complex operations without human intervention—remains elusive.
Hardware reliability is another major concern. Legged robots have many more moving parts than wheeled ones, increasing the probability of mechanical failure. In the vacuum of space, with extreme temperature swings and abrasive dust, every joint and actuator must be built to last. Redundancy, fault tolerance, and self-diagnostic capabilities are essential. Moreover, power, weight, and volume are at a premium on any spacecraft. Designing a legged robot that is both capable and compact enough to launch is a significant engineering challenge.
The review identifies four key technologies that must be advanced to make legged robots viable for deep space: sensory fusion, intelligent control, reconfigurable morphology, and multi-robot cooperation. Sensory fusion involves combining data from cameras, lidar, inertial sensors, and tactile feedback to create a comprehensive and reliable understanding of the environment. Intelligent control requires algorithms that can adapt to changing conditions, learn from experience, and make high-level decisions. Reconfigurable morphology allows robots to change their shape or configuration to suit different tasks, such as switching from walking to rolling or deploying tools. Multi-robot cooperation enables teams of robots to work together, sharing information and coordinating actions to achieve goals that a single machine could not.
The future of deep space exploration may not rely on a single type of robot, but on a diverse fleet of specialized machines. Legged robots could serve as scouts, venturing into dangerous or inaccessible areas to gather data and prepare the way for larger vehicles. They could assist astronauts during extravehicular activities, carrying tools or providing stability on uneven ground. In the long term, swarms of small, agile legged robots could conduct large-scale surveys, sample collection, or even construction tasks on the Moon or Mars.
While the technology is not yet ready for prime time, the progress made by research teams in the U.S., Europe, and China is undeniable. The prototypes described in the review represent not just engineering achievements, but visions of a future where robots move with the grace and adaptability of living creatures across alien worlds. As Sun Junkai and his colleagues conclude, the path to engineering application is clear: it requires sustained investment in fundamental research, rigorous testing in simulated environments, and a commitment to overcoming the complex interplay of hardware and software challenges. The legged robots of tomorrow may not roll, but they will walk, climb, jump, and ultimately open new frontiers in our exploration of the cosmos.
Sun Junkai, Sun Zezhou, Xin Pengfei, Liu Bin, Wei Qingqing, Yan Chuliang. Legged Robots Pave the Way for Next-Gen Deep Space Exploration. China Mechanical Engineering, 2021, 32(15): 1765-1775. DOI: 10.3969/j.issn.1004-132X.2021.15.001