My hero is an ingenious technician held captive in a pirate enclave and serving them in the mechanical maintenance of their steampunk race horses. Similar to pod racing in Star Wars I: The Phantom Menace, these races form the basis of the enclave’s illegal gambling activities one of the key sources of income as well as control of elite characters in the world. Save for basic racehorse design specs, these races are mostly unregulated. Their primary goal is high risk entertainment for the rich. But the horse-chariots themselves are very technical machinery requiring a lot of clockwork, hydraulic rams, and gyroscopically controlled actuators to make them emulate horse locomotion. A render is included below.

Chariot horse designed for illegal racing

Chariot Horse concept

(Note: wheels are not propelled in these races)

In this question, I believe my chariot-horses require some ground detection system embedded in the articulated mechanical horse hooves during fast galloping. Without some sensor system I feel they would simply topple on the slightest uneven terrain, or certainly while banking. The primary objective of these detectors therefore is to identify the ground as the hoof approaches it at high speeds, enabling the actuator machinery to prepare for weight-bearing and propelling the leg once firmly set on the ground. I have considered three different types of sensors, though other ideas might exist: sonar, gyroscopic, and pressure sensors.

Sonar sensors

utilize sound waves to detect objects and measure distances. By emitting sound pulses and analyzing the reflected signals, sonar sensors can determine the proximity of the ground. This technology has been widely used in various applications in the 20th century and today but may be difficult to justify in 19th century tech without compact electronic components, though it is not impossible since the sound is ultimately converted into electrical frequencies which can still be detected and timed even with vacuum tube technology. I think it would be cumbersome and unlikely in such a harsh environment - like putting a glass tube amplifier in a paint shaker. Its effectiveness in detecting the ground during fast galloping on a heavily vibrating mechanical horse is one option, however.

Gyroscopic sensors,

on the other hand, rely on the principles of gyroscopic precession to detect changes in orientation. By measuring rotational motion (perhaps of the shoulder beam as a leg takes load), these sensors can infer when the hoof has made contact with the ground. Gyroscopic sensors have proven to be reliable in many modern applications such as aerospace and robotics. However, their believability in the context of articulated mechanical horse hooves relying on 19th century technology is what I am concerned about.

Pressure sensors

are designed to measure force or pressure exerted on a surface. In the case of ground detection in horse hooves, pressure sensors can be embedded within the hoof structure to detect when it comes into contact with the ground. These sensors can be linked to hydraulic gates sending information about weight distribution and timing for actuator machinery preparation. This would require hydraulic sending lines to reach the hoof, where they will be subjected to constant jarring, vibration, bending, and shock. Possibly I am overthinking this, and a sensor in the head would be adequate to inform the actuator machinery about the distance to the ground. In that case the head position and level needs to be very accurately measured at each step, and any flexibility in the neck or frame will throw off the measurements. Not impossible, but very difficult.

To solve the believability problem of these vehicles, the story needs a believable technology that allows a thing like this to stay upright but does not overcomplicate the machine. It is still a race horse after all, and weight is one of the considerations.

A good answer suggests the best fitting tech for explaining galloping mechanical horses, identifying strengths and weaknesses over each alternative detection method.

The answer will contribute to the suspension of disbelief about having efficient and reliable ground detection systems in articulated mechanical horse hooves in this steampunk setting. With an understanding of the capabilities and limitations of sonar, gyroscopic, and pressure sensors that modern readers may have, the audience can hopefully just accept this aspect of the story as the plot unfolds around it. I will reiterate that these races are illegal and many unfair practices are a part of the sport, very much like the pod races in Phantom Menace. This means that the more complex the solution is, the less likely it is to succeed on the track against a competitor. So there is a balance between not enough and too much engineering here.


2 Answers 2


Frame Challenge

Given your Render, the Chariot wheels are providing most of your stability, so you probably don't even need any sensors as such.

I think a simple pressure plate type system in each 'hoof' that is set to trip at a certain pressure (say 100Kg) that is then mechanically linked to to the pistons that drive the leg, similar in function to how a Slide valve is linked to the driving rods of a steam locomotive and controls the movement of the piston.

The reason for this is so that each hoof is planted with enough weight on it (and in solid enough ground) before it is actuated. You could even get a little fancy and add in a Govenor type setup that as the 'horses' speed increases, the gait and timing of the 'hooves' is changed - e.g. from a walk, to a trot, to a canter then to a full blown Gallop - as at a Gallop, the Horse is airborne for a period of time.

I feel this would make it more believable and SteamPunk-y.

  • $\begingroup$ I mentioned in the W that possibly none was needed however the center of gravity is very high. I don’t see this banking well if it were rigid. $\endgroup$
    – Vogon Poet
    Commented Aug 18, 2023 at 4:04
  • $\begingroup$ @VogonPoet - for banking/cornering, you could have an elegant mechanical linkage type system to lean the horse into the corner, even better if this is controlled by the driver as it adds an element of skill to the race (and therefore, better betting opportunities) $\endgroup$ Commented Aug 18, 2023 at 4:19
  • $\begingroup$ The rider (driver) certainly has the reigns, these machines can’t sense lanes or turns. The hip pivots to shift weight. But this offsets the stroke and stride of each foreleg, so fixed-length strokes would make for rough turns. $\endgroup$
    – Vogon Poet
    Commented Aug 18, 2023 at 4:31
  • $\begingroup$ @VogonPoet - Then just have it that as it pivots, it mechanically adjusts the stride/stroke. This would be reasonably easy to do and although it wouldn't be perfect, a set ratio would be good enough for the majority of situations. $\endgroup$ Commented Aug 18, 2023 at 9:01
  • $\begingroup$ I’m thinking it would have jerky torque like the old positive traction vs. limited slip differential unit problems, or old style 4WD turning on pavement. A slight incline or decline, or a divot would amplify the differences in stride. So a perfectly engineered track would allow a simpler design, I think the racers will be forced to adapt to variables and sub-par conditions. Even with a good track, the adaptable stride would trade some weight for a performance advantage. It’s an option tho. $\endgroup$
    – Vogon Poet
    Commented Aug 18, 2023 at 11:00

Real horses can do this with a limited set of sensors. You can blindfold a horse, and as long as the ground does not change unpredictably, a horse can continue to run.

The relevant senses that horses have are:

  • Force sensors, which can gauge the force being generated by muscles.
  • Positional sensors, to tell what position the parts of their bodies are in.
  • Balance sensors, to tell the overall position of the body in their environment
  • Touch sensors, so that they can feel when their hooves touch the ground.

Some of these senses overlap with one-another.

Now, if your robot horses are running around a flat circuit, and are under the control of their riders/drivers for speed and direction, that's pretty much all they need. Slight irregularities in the ground surface are not going to make a blind horse fall over, given all of its non-visual senses, and the OP's robots need not fall in such conditions either.

However, if these robot horses will need to be used on uneven ground, then they will also need some sort of ranging sense so that they can detect and negotiate obstacles and uneven terrain. Horses use sight, but sonar might work just as well or even better, since sonar is more of a ranging sense than sight, and visual processing is a more complex task.

However, the existence of modern autonomous quadrupedal robots shows that this is not an insurmountable problem.


Since the OP now wants mechanisms to be discussed...

Each gait cycle effectively has a limb cycling at a given speed, in a given sequence between all limbs. Surface detection by touch can be used in combination with force sensors, proprioception and balance sensors to apply the appropriate upwards forces to keep the automaton's body level. When the body is level, the limb would apply more propulsive effort than lifting effort. The limb root being too high would result in less lifting effort, while the limb root being too low would result in more lifting effort beiong applied. This is simply a negative feedback loop.

Where an uneven surface exists, rangefinders could be used to cause the limb to extend more or less in order to avoid an obstacle or reach a lower surface. Again, this is a negative feedback loop, albeit a slightly more complicated one.

  • $\begingroup$ An elegant answer, how are inclines and declines handled? It seems like the level balance will be trimmed by the driver to anticipate grade changes. Or was that what you meant by range finders? Is this a sonar / pressure plate hybrid? I’m expecting the drivers to have their hands full shooting harpoons and raising shields, so as much automation that is believable and durable is best. Engineered like a Battlebot - lots of variables, driving skill goes last. $\endgroup$
    – Vogon Poet
    Commented Aug 22, 2023 at 3:02
  • $\begingroup$ @VogonPoet The rangefinder would be responsible for providing data for limb extension/retraction in the case of downwards and upwards gradients respectively. Going up or down a slope would cause a delayed alteration of the automaton's balance, which would allow for changing slopes without forcing continual overextension or underextension. $\endgroup$
    – Monty Wild
    Commented Aug 22, 2023 at 3:33

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .