How to build a high speed rover?

[Matrix Aran]

So, as the title suggests, my question is about the physics of building and tuning a rover to manage to reach the highest posssible speeds over a variety of terrain.

After doing some searching I managed to find a few of the old racing challenge threads but many were before the latest changes to rover wheels and few posts went into any of the details or logic of their designs, at least not at the level of detail I'm interested in. I've also gone searching for high speed rover tutorials but I've not found many.

It's all well and good that a rover can reach its maximum speed on a flat runway, but I'd like to be able to tune my rover to achive the greatest possible speed over terrain, while manuvering. I realise, much like with real world racing, that there's a limit to how fast you can turn a wheeled vehicle and how bumpy the terrain can be, but I'd like to get my designs as close to that limit as I can, and I'm hoping you kind folks can help me do that.

As for the body where the rover will be, I'd Ideally like three design strategies. One for bodies that have a very low surface gravity, such as minmus, another for bodies with more modest gravity, such as Duna, and a third for the Eve/Tylo classes of planet. Given that rovers usually cross one biome to the next, they'd have to deal with hill climbing/descents, and obstacle avoidance of the new surface features.

[ARS]

Make your rover with high ground clearance, and very low center of mass (IIRC, Rapier engine has a bug where it's center of mass is offset forward. This can be used to manipulate the CoM, but I don't know this is patched or not) and set reaction wheel to SAS only. Ideally, for low gravity world (especially without atmosphere), an upward facing low-power RCS engine is needed to maintain ground force to prevent your rover rolls into orbit at high speed driving. Do not make a sharp turn at high speed. Pack a lot of batteries for long endurance and supplement it with solar panels or RTGs (Your batteries only drains if you accelerating). You can add small thrusters as nitro boost when going over incline and for aerodynamics, you can throw it out of the window on airless celestial bodies since it's pointless there

Edited July 20, 2019 by ARS

[Matrix Aran]

 

3 hours ago, ARS said:

Make your rover with high ground clearance, and very low center of mass (IIRC, Rapier engine has a bug where it's center of mass is offset forward. This can be used to manipulate the CoM, but I don't know this is patched or not) and set reaction wheel to SAS only. Ideally, for low gravity world (especially without atmosphere), an upward facing low-power RCS engine is needed to maintain ground force to prevent your rover rolls into orbit at high speed driving. Do not make a sharp turn at high speed. Pack a lot of batteries for long endurance and supplement it with solar panels or RTGs (Your batteries only drains if you accelerating). You can add small thrusters as nitro boost when going over incline and for aerodynamics, you can throw it out of the window on airless celestial bodies since it's pointless there

So what you've laid out is the basis of most of my current rover designs. Low CoM, wide wheelbase, with an SAS only reaction wheel. 

The problem is I don't find that good enough, and I still have more questions. What pris and cons are there to more than four wheels on the rover? Can the suspension be tuned so that small dips in the terrain don't launch the rover into the air? Any way to keep the wheels from locking up on landing?

I'm looking for the details that could offer better than just the baseline.

[Rocket In My Pocket]

Well, my approach typically is just to wrap the whole thing in 80m/s resistant structural pieces to form a roll cage.

Who cares if it rolls or launches into the air if it's indestructible?

Spoiler

Exploiting the COM of any jet engine works as mentioned above to push COM lower, even below the ground if you like. Many of my rovers include this to some degree.

Another interesting concept is "pseudo hover crafts" with wheels set to zero friction. Can turn at high speeds, and navigate a lot of uneven terrain, but prone to random unpredictable accidents/explosions.

Spoiler

One more interesting design to give you some ideas, this one was an attempt to passively mitigate harsh rolling hills and inclines.

Spoiler

 

Edited July 21, 2019 by Rocket In My Pocket

[Matrix Aran]

12 minutes ago, Rocket In My Pocket said:

Well, my approach typically is just to wrap the whole thing in 80m/s resistant structural pieces to form a roll cage.

Who cares if it rolls or launches into the air if it's indestructible?

  Reveal hidden contents

Exploiting the COM of any jet engine works as mentioned above to push COM lower, even below the ground if you like.

Another interesting concept is "pseudo hover crafts" with wheels set to zero friction. Can turn at high speeds, and navigate a lot of uneven terrain, but prone to random unpredictable accidents/explosions.

  Reveal hidden contents

 

I am interested in using wheel friction to tune a rover to understeer at speed and avoid flips, but I'd rather not induce any other sillyness/instability.

Ridiculously strong structural parts and engine CoM sillyness aren't quite satisfying, as they somewhat trivialize the engineering of a rover, though I do appreciate them being mentioned as I'm sure other readers of this thread may find a use for them.

[mystifeid]

Each planet/moon type really needs a different design and a design will really depend on how far you intend to travel around that planet/moon. Driving around close to KSC will produce a design different to one used to circumnavigate Kerbin. I was interested in the latter and also very interested in finding the best designs for speed.

 

This is what I used drive around Kerbin in a pretty straight line polar route.

Has a panther for high top speed on the flat. Uses separately tunable control surfaces to optimize ground contact at speed.

Upside down, the canards act as hydrofoils (their main purpose) giving 300+km range in the water at 60m/s.

Can climb near vertical slopes even transitioning from the water. Contains everything necessary to mine and refuel.

The TR-2L wheels are 100m/s impact parts and driving at anything around these speeds is inherently dangerous. Loss of ground contact will always happen and then it is up to you to land the rover with as many wheels as possible touching down at the same time. Having more wheels seems to help spread out the impact allowing higher jumps and harder impacts. Increasing spring and damper strength also seemed to help at the time (except on Eve). Sharp changes in uphill inclination will also occur and it is up to you to anticipate these and reduce speed in order to avoid exploding the front wheels. Tylo and Vall are probably the places where it's possible to maintain the highest speeds without using engines due to lack of atmosphere, reasonable gravity and a lot of long gentle downhill slopes. This is the fastest I dared allow my 16 wheeler to go when driving (downhill) on Tylo.

Eve is by far the most challenging place to circumnavigate. Just getting the rover to the surface is hard. Travelling around Eve in a reasonably straight line is possible without getting your feet wet by taking a polar route. Power and not speed is what is important here. I used action groups to toggle power on/off to pairs of wheels to conserve ec or to provide more grunt. Each double wheel pair of this 20 wheel rover was tuned differently and I still must have had 100 blowouts, mostly in any biome higher than the Lowlands. On Eve, 40m/s is about the speed where you start to seriously risk the lives of your Kerbals.

Highest point reached. This rover can climb pretty steep slopes on Eve.

Close up.

Fixing yet another blowout

Fairly typical polar inclination change of 45° or so. Like driving into a wall. Power and care required.

For myself, I think it's better to learn how to deal with jumps and barrel rolls rather than trying to totally negate them. Things like downward pointing engines can produce their own problems. When I first started I used to hyperedit a rover to the Mun and use Mechjeb to set a straight line course and speed of 30-40m/s then sit back and wait for something to break. Then build a stronger rover. By the time the rover is landing jumps from 150m up or more onto flat ground, it's probably in the ballpark. And it helps to see how Mechjeb lands jumps.

The picture below should give some idea of the difference in size of the rovers I used for my cn's. Just because you suddenly find yourself in a barrel roll 10m above the deck doesn't mean all is lost. On Tylo alone I probably successfully recovered 10-20 barrel rolls in the big rover which is very wide with a very low COM. That is, with SAS on :
- roll to stop the rotation
- yaw to align in prograde direction
- roll and pitch to ensure all wheels touch down simultaneously.

To do much rovering at all, I think it helps to get the practice to be able to do this automatically. That way you tend to become a little less hung up about the design.

 

Edited July 22, 2019 by mystifeid

[Snark]

On 7/20/2019 at 1:00 PM, Matrix Aran said:

with an SAS only reaction wheel.

I'm curious, why SAS only? Doesn't that mostly defeat the purpose of having the reaction wheel?

After all, in SAS only mode, the reaction wheel is useless for reorienting yourself to land properly when you go airborne-- you'll lose all control the moment your wheels lose contact with the ground. And you won't be able to use it to assist with turns, or as a proactive anti-tumble compensation for when you need to do a hard maneuver at speed.

[Matrix Aran]

23 minutes ago, Snark said:

I'm curious, why SAS only? Doesn't that mostly defeat the purpose of having the reaction wheel?

After all, in SAS only mode, the reaction wheel is useless for reorienting yourself to land properly when you go airborne-- you'll lose all control the moment your wheels lose contact with the ground. And you won't be able to use it to assist with turns, or as a proactive anti-tumble compensation for when you need to do a hard maneuver at speed.

The main reason is that I've been far (far) too lazy to rebind the default rover controls from WASD, so when its in SAS only mode, the WASD keys work as normal. If I go so airborne that I can swap it over to pilot mode, I'll re orientate, but most of the time I'll only let myself get airborne if I think the launch angle will also land well. This also kinda factors back into why I'm so insistent on trying to tune a design to not be unstable in the first place. I want to go fast, but I don't want to get -too- cheesy with the SAS wheels.

Another factor in asking this question is that I keep figuring that there must be a better way to design a rover that can go fast. The engineering problems likely have cropped up in the real world, but I don't have that much automotive engineering experience to have any sort of intuition about how to further refine a design.

[mystifeid]

21 minutes ago, Matrix Aran said:

but most of the time I'll only let myself get airborne if I think the launch angle will also land well

So that would take 90% of the fun out of it for me. It doesn't really make sense to me that you are unwilling to spend a minute to rebind the rover keys yet are happy to spend how long ? to do something that imho can't be done. And just because a rover is stable on Kerbin ... my big rovers are impossible to flip on the flat on Kerbin when using wheel power but on Vall for example, that is far from the case.

Anyway, good luck with it.

[mystifeid]

Although this challenge closed a long time ago, you could still use the race section of the Dessert Derby as a yardstick for what 'fast' is. The course basically just has a start line near the Dessert runway and a finish at the temple with the course taking in varied terrain (and it looks like the savefile is still available for download). Possibly, due to limited participation, we're still not seeing what 'really fast' is, but it may help as a useful test. (And for the record, in the race section, I'm happy finishing second to Katateochi any day of the week)

[Matrix Aran]

7 hours ago, mystifeid said:

So that would take 90% of the fun out of it for me. It doesn't really make sense to me that you are unwilling to spend a minute to rebind the rover keys yet are happy to spend how long ? to do something that imho can't be done. And just because a rover is stable on Kerbin ... my big rovers are impossible to flip on the flat on Kerbin when using wheel power but on Vall for example, that is far from the case.

Anyway, good luck with it.

We all have our own ideas of fun

I see it as an engineering challenge, and I agree it probably can't be done, simply because I doubt there's a hard limit where there's no way to squeeze a bit more out of the engineering.