Watching Boston Dynamics progress with their walking robots makes it seem like maybe a version adapted to function off-world could compete with wheeled rovers now:

This looks like a machine that could handle a lot of terrain a wheeled rover can't, and could be more effective with tools as it is capable of bracing itself, and navigating fluidly around something it is investigating or working on. It doesn't seem like such a big step to make legs that can function also as arms at least in a limited way, either.

On the other hand, maybe the mechanisms involved are too delicate. It had occurred to me that maybe joints that aren't rotating like wheels are easier to shield from dust, and have an advantage because the motors are above the dust, and the feet don't kick up dust like a wheel can. But I could be off-base on that. Maybe their power requirements are a big disadvantage, though I haven't found data on how much power they draw.

Has it now become feasible to adapt this technology for a walking robot on another world? (I am aware it hasn't been developed and tested for that application, but I'm referring to taking what there is and adapting it.) Does the format have the advantages it seems to have, or does the complexity and fragility of the technology still outweigh possible advantages? I am particularly interested in how they might perform on the Moon, with its nasty dust, and the traction issues that come from low gravity. As my focus is looking at what extensive permanent infrastructure might be like, power usage and supply in that light is my interest.

In this video, another version handles some more varied and challenging terrain:

  • 2
    $\begingroup$ If you're interested in terrain-handling, note that the closest thing to "irregular terrain" seen in that video is a staircase. I'll be impressed when someone shows one of these things walking across a talus slope. $\endgroup$
    – Mark
    Aug 2, 2016 at 1:47
  • $\begingroup$ @Mark i added a video where Big Dog handles some more difficult terrain, though admittedly a talus slope is a lot more relevant. $\endgroup$
    – kim holder
    Aug 2, 2016 at 2:28
  • 2
    $\begingroup$ Incidentally, the Soviet Mars-2 and Mars-3 landers came with a sort of walking rover. The rover was on an umbilical and walked on a pair of skids, alternately lifting them while resting on its body, moving them forward, planting them, and pushing them back to advance the body, like a child's toy. Both landers failed before deploying their rovers, sadly. $\endgroup$ Aug 2, 2016 at 2:37
  • 1
    $\begingroup$ Have a look at ATHLETE (All-Terrain Hex-Limbed Extra-Terrestrial Explorer). JPL has put quite some thought into this already. One drawback is that legs generally are heavier than wheels. ATHLETE compensates for this by using the legs/arms for more purposes than mobility. (Your videos are creepy!) $\endgroup$
    – LocalFluff
    Aug 2, 2016 at 6:41
  • 3
    $\begingroup$ This system is more complex: more possible failure points. One one leg fail it's hard to recover. Whereas one or two failed wheels can be mitigated. $\endgroup$
    – Antzi
    Aug 2, 2016 at 16:54

3 Answers 3


From a first principles point of view, to move around on a rough surface with a reliable robotic vehicle for extended periods of time, it's probably unavoidable that you need to expend energy. Walking 10km on gravel is more work than a flat surface because the gravel moves and dissipates energy. Walking 10km over rocks that don't move still uses energy because you have to go up and down doing work against gravity and we don't recover that energy.

enter image description here

above: Plot of some randomly selected potentially interesting solar system bodies. Horizontal axis: surface gravity (somewhat related to energy needed to move around) as a ratio to that on Mars. Vertical axis: Approximate sunlight intensity as a ratio to that on Mars - as estimated by ratios of semi-major axis to the -2 power. Venus is listed twice - at the top of the atmosphere where aircraft such as robot balloons and robot planes can collect substantial amounts of light, and the surface where only a few percent of the redder parts of sunlight reach.

enter image description here

above: Page 6 from Venus Aircraft - design evolution 2000-2008, Geoffrey A. Landis, NASA John Glenn Research Center. Above 50km, there is more sunlight available than there is on earth - closer to the sun, and possibility to collect reflected light from below (as does the ISS around Earth) to make up for some cloud cover.

enter image description here

above: Page 32 from Venus Aircraft - design evolution 2000-2008, Geoffrey A. Landis, NASA John Glenn Research Center. The very dense atmosphere makes powered flight very attractive (ballooning as well). However, flying as fast as the wind would be energetically challenging at most altitudes.

Robotic Areal Vehicles may be possible future missions for Venus' atmosphere. It's a long reach, but things taking inspiration from the Festo Air Penguin discussed at length in this answer and shown below, and the Festo Air Ray (not shown) might be possible.

above: Festo Air Penguin discussed more here.

above: Festo Robot Balloon delivering a bottle of water on demand. This would be much more difficult on Venus for a number of reasons, but the higher atmospheric density means balloons could carry a significantly heavier payload, and it wouldn't be necessary to use Helium for buoyancy.

above: Festo Bionicopter could take advantage of the denser atmosphere on Venus. It could also make use of some legs as well!

More about the Vega program using robot balloons on Venus in Wikipedia, in Wired, and in The New Scientist, and future possibilities with NASA's Venus Exploration Group (VEXAG) and ESA's European Venus Explorer (EVE).

Wheels have served humans well over thousands of years. Through zillions of km of trial and error as well as amazing engineering, they’ve solved mobility problems for humans here on earth and on several other solar system bodies.

By far the largest body of detailed experience, imagery and metrology of wheel performance on off-world robotic vehicles comes from the three rovers on Mars.

enter image description here

above: Comparison of Mars Rover Wheels. Left: Sojourner of Mars Pathfinder mission. Center: Mars Exploration Rovers (MER) (Spirit and Opportunity). Right: Curiosity of the Mars Science Laboratory (MSL).

enter image description here

above: Curiosity Self Portrait at Big Sky Drilling Site.

enter image description here

above: Detail cropped from Curiosity Self Portrait at Big Sky Drilling Site.

enter image description here

above: "Location map - Curiosity rover at the base of Mount Sharp - as viewed from Space (MRO; HiRISE; February 4, 2016/Sol 1243)." You have to open this into a separate window and zoom in to see the trail details, starting in the right side of the upper edge. Note that the path is chosen as a compromise between science and where the wheels are judged safe enough to go without getting stuck or damaging the vehicle.

One of the jobs of Curiosity's Mobility System is to cary a large package of Curiosity's Scientific Instruments over large distances so that information can be collected from a wide variety of locations.

enter image description here

above: Curiosity Robotic Arm applying a drill to Martian rock. Samples are then collected and transported to locations inside Curiosity for further analysis using a variety of analytical equipment.

enter image description here

above: Curiosity still ...inside the Spacecraft Assembly Facility at NASA's Jet Propulsion Laboratory, Pasadena, Calif.. Even in Mars's surface gravity only $\frac{3}{8}$ of Earth's, all of this scientific equipment together has gotta be pretty heavy! The robot arm is often forgotten because it doesn't show up in many of Curiosity's selfies in the same way your hand or the business-end of your selfie stick don't show up. But if you look close in the Big Sky Drilling Site "selfie" a few images above - you can see it's shadow on the surface!!

While Boston Dynamics' Big Dog ran on fossil fuel for various reasons (see this Boston Dynamics conference proceeding PDF) including power density and demonstrations for particular non-science "missions", Spot, SpotMini (shown in the question), and LittleDog are electrically powered, and LittleDog seems to be built with off-world use in mind, or at least in the back of the mind.

LittleDog has four legs, each powered by three electric motors. The legs have a large range of motion. The robot is strong enough for climbing and dynamic locomotion gaits. The onboard PC-level computer does sensing, actuator control and communications. LittleDog's sensors measure joint angles, motor currents, body orientation and foot/ground contact. Control programs access the robot through the Boston Dynamics Robot API. Onboard lithium polymer batteries allow for 30 minutes of continuous operation without recharging. Wireless communications and data logging support remote operation and data analysis. LittleDog development is funded by the DARPA Information Processing Technology Office.

enter image description here

above: Little Dog cut-away illustration from Boston Dynamics

above: Video of Little Dog climbing over terrain from here.

above: Video of Boston Dynamics' battery-powered Spot climbing over terrain and getting along well with "Hockey Stick Guy" (here with commentary on YouTube and in Wired) despite getting kicked by him.

Presumably, a main robot rover may also cary one or several more highly mobile rovers for sample collection. Much in the same way that Curiosity's robotic arm can collect samples and transport them to the "laboratory" inside Curiosity, mini-rovers may be able climb up on to, as well as down into hard-to-reach places for measurements, imagery, mapping (via telemetry or data transfer upon return) as well as some kinds of sample collection. While drilling requires force and Curiosity uses mass and leverage, a clever robot could find leverage between rocks or walls, possibly even moving rocks around to improve the situation.

These guys look like they are ready to go anywhere in the solar system!

above: Boston Dynamic's Sand Flea launching itself all over the place! Now imagine this happening on a low surface gravity body. Suborbital (except in extreme cases like comets or small asteroids) but it is point-A to point-B transportation. Needs robust electronics and sensors to avoid getting a headache, but possibly fine for sample collection and scouting.

Currently it uses stored compressed gas for multiple jumps (see below). Some interesting ideas could be imagined to make the gas rechargeable from an atmosphere, or replaced with an electromagnetic linear motor (tiny captive rail-gun-like thing).

The following is from the Sand Flea Datasheet (remember, the specs are for Earth surface gravity!):

SandFlea is a small robot with remarkable mobility. The 11 lb robot drives like a traditional wheeled vehicle on mild terrain, but jumps up to 8m high over difficult terrain. It can jump 25 times using the piston actuator and onboard fuel supply. Jumps of 1-8 m heights are user selectable. Specially designed wheels cushion the shock of landing. Flight and landing attitude of the robot are automatically controlled by an onboard stability system.

  • Controllable hop height, 1-8 m
  • Controllable launch angle
  • Precision hops through windows or doors, on to tables, up staircases, on to or off of roofs or balconies
  • Piston actuator
  • Laser-based ranging to guide launch
  • Operator control unit (OCU) with live video feed for remote operation
  • Robot and OCU both fit into a small backpack

enter image description here

above: Boston Dynamics says:

The robot uses gyro stabilization to stay level during flight, to provide a clear view from the onboard camera, and to ensure a smooth landing.

above: Boston Dynamic's RHex going all over the place - looking for water perhaps?

However, they would either need tiny RTGs of their own, solar panels of their own, or have to be charged and then recharged by the main robot. This can be done by contact, or through highly resonant inductive charging - which can actually cover a significant gap of a few meters in a pinch - or just optical charging - laser to special photovoltaics like this:

above: quadcopter illustrative example of a small vehicle receiving power from a beam of light. Note: an aperture of 5 centimeters can "beam" power over many kilometers if atmospheric effects are minimal and motion is minimal.

  • $\begingroup$ Perhaps worth mentioning that experience with off-Earth surface mobility goes back to Apollo with the Lunar rover, which had to fit within a very restricted mass budget; it was battery powered (non-rechargeable) and had to carry two space-suited astronauts and their gear, and had a range of 57 miles. One factor in its favor as compared to mobility on Mars is the lower Moon surface gravity. The better rough terrain performance of a walker would have to be traded off against its added complexity and presumably, mass and power requirement. $\endgroup$
    – Anthony X
    Aug 7, 2016 at 15:31
  • $\begingroup$ @AnthonyX Thanks! I'm not doing a survey of history, or I'd talk about the Soviet lunar rover and the Chinese lunar rover. That would be a good question but someone else would have to answer. If you want to go from point A to point B and you must cross rocky terrain to get there, I am not sure a wheel-based solution uses less energy than a leg based solution. Wheels use less energy only if you limit yourself to strictly wheel-friendly flat terrain. Complexity is always a problem with everything until after you do it, when it becomes standard. Rovers are crazy-complex, a few more motors - eh! $\endgroup$
    – uhoh
    Aug 7, 2016 at 15:41
  • $\begingroup$ @AnthonyX Curiosity has six wheels, each has a drive motor and four of them (?) have steering motors, that's ten. Little Dog has 12. Where's the difference in complexity? Future computers and decision making algorithms will support calculations for leg motion planning as well as computers of the past are supporting rovers. Back to the first point - look at the "Location Map" and link of curiosity I included above. That is not the most interesting path to science - it's a serious compromise based on wheel-friendly terrain only. $\endgroup$
    – uhoh
    Aug 7, 2016 at 15:43
  • 1
    $\begingroup$ @WaterMolecule I see what you mean. This was five years ago and I think I've improved a bit in post-writing since then. I had plans to keep working on this; I'd gone to the library and did a lot more research (and downloading) on legged-robot research (there were a lot of spider-like robot projects in previous decades) but ultimately lost steam and left this as-is in its suboptimal state. $\endgroup$
    – uhoh
    May 24, 2021 at 20:33
  • 1
    $\begingroup$ @uhoh Yeah, sorry, I just realized that this post is 5 years old and that I could probably improve it myself. I upvoted your new answer before I even saw this one. $\endgroup$ May 24, 2021 at 20:36

One possible reason may be the Technology Readiness Level of these walking machines, considering both the mechanic and the supporting electronic: wheels solution can be considered "good enough" for the current mission's objectives, and have the best reliability for operating in that environment.


Thanks to the useful comments, I realized I gave a very poor supported answer, so I will draw the economical model behind my one-line.


In the graph you can read on the y-axis a variable, "Level Of Technology required for the mission scope (LOT)", that can be thought as the sum of, as example, reliability for the duration of the mission, sturdiness necessary for the explorations planned, energy consumption within mission constraints, and so on.

On the y-axis we place the year. Different lines are different technologies: I plotted wheel-based rovers, leg-movement based robots and, to project further the analysis, an hoverboard-like drone.

For every planet exploration mission, the LOT is almost constant: there can be minimal variation in case we want the rover travel on flat terrain or climb a mountain (Mars) or navigate over rather than under ice (Europa), but the main parameters are terrain composition at the rover-size level (density of the soil, operational temperature, and so on). For my model, I consider them constant.

Model is not in scale, since variable are just a raw approximation of reality, but I consider Mars more challenging that the moon for some known characteristics already explored by rovers and satellites, and Mercury and Europa even more challenging just because of the mission constraint to actually go there (distance) and geology (ice surfaces, extreme temperature ranges).

The interception between the wheel-based exploration and Moon LOT represent Apollo missions, while interception with Mars LOT represent martian rovers. If we project this line in the future, we may have advances in technology that will allow wheels also on Mercury and on Europa.

On the other line, leg-based robot are on an unknown trajectory: we can consider them adequate for Earth exploration but not yet for other planets mission. Same for hoverboards: we have even more unknowns. But depending of the speed of advances of these technologies, we may have one or the other reach an intersection point with the wheel-based approach, and in that case there won't be any rational cause to not consider them the "best" option available.

As extreme, the hoverboard line is vertical, as we may think that hoverboard will travel on every condition and terrain, therefore they will be the best choice as soon as they're introduced.

We may want to populate the model with as many parameters as we can get (e.g. more funding expected for the leg-based technology can tilt up the leg-based line, while new materials to increase energy efficiency may tilt all the lines), but I think we should face that the situation now is this:


The amount of info required to produced a precise model is simply too much, and I think a clear answer is premature to give without falling in speculation: we simply do not yet have enough data.

(model derived from Clayton Christensen, "The Innovator's Dilemma")

  • $\begingroup$ I Like the second graph. It's kind of honest. Legs did work well in evolution. Even birds and flies and pilots have legs. $\endgroup$
    – LocalFluff
    Aug 10, 2016 at 18:46


All of these (extra) degrees of freedom mean that the (articulated) wheels can do things that the Mars rovers, with their rocker bogie suspensions, just can’t.

How about legs with wheels on them performing a combination of rolling and a bit of walking at the same time?

Wiggly Wheels Could Help Keep Rovers From Dying on Mars; Unconventional gaits keep this rover from getting stuck in tough terrain

That full-size RP15 rover that you briefly saw in the video was a test bed for the Resource Prospector rover, which was going to do some exploring of potentially icy parts of Earth’s moon in 2022. Resource Prospector was canceled in 2018, and then some of what it was going to do was resurrected as part of the VIPER lunar rover. But VIPER’s wheel system is quite different. The RP15 prototype uses a four-wheel design, but the wheels are on a sort of actuated suspension system that allows them to move up and down and forward and backward as well as rotating around the vertical axis. All of these degrees of freedom mean that the wheels can do things that the Mars rovers, with their rocker bogie suspensions, just can’t.

We first investigated an open-loop gait derived from tests of RP15’s crawling capabilities at JSC. Previous studies showed that various open-loop strategies for granular slope climbing were sufficient if the locomotor’s dynamics allowed it to repeatedly intrude into undisturbed media. We implemented this gait on the Mini Rover by cyclically sweeping rearward with three appendages while one appendage lifted to disengage from the medium and also spinning all four wheels at a constant rate of 2.1 rad/s. This gait is classified as a quadrupedal rotary sequence (RS) gait with regard to its foot placement, which cycles around the rover’s locomotion appendages.

There's a lot more in the article, the block quotes serve only to demonstrate that an established "gait" is part of the way this curious "walking wheel" implementation works.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.