Northward autonomous rover challenge!

[Pds314]

The goal of this challenge is to design, build, and run a fully automated ground vehicle which, starting from the KSC runway or launchpad, gets as far north as possible in latitude before getting stuck or lost.

Rules:

1. You may not manually control the vehicle in any manner after launch. It should be completely "Fire and Forget."

2. It's ok to manually mess with SAS, trim, staging, action groups, movement, SAS modes, part settings, etc, ONLY BEFORE launch. A vehicle will be considered launched once it leaves the runway or launchpad structure.

3. The vehicle should be on the ground, not in the air or in space or on the water's surface. It's ok if it makes short jumps or crosses some water but it's not ok to have it fly or boat most of the way. It should be on land for the supermajority of the distance.

4. You may launch as many vehicles as you wish in any sequence, but all must be automated and follow the rules.

5. KAL, if used, must only modify values within their legitimate slider range. No overclocking or underclocking.

6. The mission is scored from the highest latitude the vehicle reaches. You don't have to wait for it to actually die.

7. If for some reason a vehicle manages to get very close to the North pole (above 89 degrees latitude), the tiebreaker is the time to get there, including all launches in that run.

 

Edited June 26, 2022 by Pds314

[Pds314]

First attempt:
The wheels are set to friction control of 0.01 and the motor is set to 1% power.
The KAL is set to just turn the motor on, then activate the yaw control 5 minutes later. I had SAS set to point normal, which is due north. A very crude approach, but it was good enough to reach a latitude of 6.127688 degrees north within 48 minutes and 41 seconds, before running into a moderately steep mountainside and flipping.

Runway:


Unguided climb out of the KSC greens:


Cruising unguided west of the KSC, we need to avoid dumping it in the water.


The KAL has activated yaw control on the rudders.


Passing by a piece of Breaking Ground collideable groundscatter. This stuff is slightly scary just because we could eventually run into one.

We've reached 1 degree north:


The rover begins to veer east due to the hillside. This is a mixed blessing. On the one hand, we don't have to go over the hill that way. On the other, the rover is highly susceptible to drifting off course and could even end up depositing itself somewhere dangerous because of this tendency.

2 degrees north. Note the intimidating slopes over there.

The rover is in fact too intimidated by the slopes and decides to veer to the east again.


3 degrees north.


4 degrees north. We can see a mountain directly ahead.


5 degrees north.


6 degrees north. We come dangerously close to another Breaking Ground tree.

The rover begins veering west climbing the hill.


And starts going back down the hill. This is the furthest we got.


The rover tumbles.


And lies flat on its back, immobilizing it permanently.

[Pds314]

Made it more stable on the ground, less stable aerodynamically, and gave it a docking port on a servo so I can aim it at any heading I want. Currently, I have it set to take ten minutes for the initial turn, and then just go due north, but in principle I can program in all sorts of course changes at predetermined times.

Under these conditions, it made it 13.458724 degrees North. It was about to run out of power when it found a really sharp hill and jumped and flipped.

Going by the mountain range I lost to before. It no longer flips.


Rover flips right before running out of charge.


As it's flipped, it's no longer recoverable.

[Pds314]

Not an entry yet but I did think this was pretty impressive. It survived so many jumps.

Edited June 26, 2022 by Pds314

[18Watt]

Working on an entry.  Nothing fancy, just point north and let it go.

I used a bunch of RTGs, so electricity isn't a problem.  I should have launched closer to morning though, so I could see where I'm going.  Or maybe put lights on it.

Just passed 6N, hit a small mountain head-on.  Seems to be handling it so far.

I considered finally figuring out how to use a KAL, but the KAL doesn't control anything I want it to- such as wheel steering or maintaining a constant rover speed.  So I didn't use a KAL.  My rover is amphibious, so I didn't need to figure out how to drive around the bay to the north of KSC.  

I'll post an actual entry when this one runs out of steam.

Edit:  Just ran out of steam, couldn't make it up the mountain.  6 deg 24' N.

Edited June 27, 2022 by 18Watt

[Pds314]

4 hours ago, 18Watt said:

Working on an entry.  Nothing fancy, just point north and let it go.

I used a bunch of RTGs, so electricity isn't a problem.  I should have launched closer to morning though, so I could see where I'm going.  Or maybe put lights on it.

Just passed 6N, hit a small mountain head-on.  Seems to be handling it so far.

I considered finally figuring out how to use a KAL, but the KAL doesn't control anything I want it to- such as wheel steering or maintaining a constant rover speed.  So I didn't use a KAL.  My rover is amphibious, so I didn't need to figure out how to drive around the bay to the north of KSC.  

I'll post an actual entry when this one runs out of steam.

Edit:  Just ran out of steam, couldn't make it up the mountain.  6 deg 24' N.

Probably the exact same mountain as mine had issues with.

4 hours ago, 18Watt said:

I considered finally figuring out how to use a KAL, but the KAL doesn't control anything I want it to- such as wheel steering or maintaining a constant rover speed.  So I didn't use a KAL.  My rover is amphibious, so I didn't need to figure out how to drive around the bay to the north of KSC. 

Yeah about the only way to control wheel steering with KAL is to attach the wheel to something you can control, e.g. robotics parts, and steer it that way, or steer it indirectly using control surface collision.

Though to be fair, control surface collision isn't a half bad way to do that. I might actually try that since it lets SAS manipulate the wheels, and you can do it in a way that doesn't produce problematic aero forces (or if you really way to, no aero forces at all).

Edited June 27, 2022 by Pds314

[QF9E]

Is kOS allowed?

Edited June 27, 2022 by QF9E

[Pds314]

11 hours ago, QF9E said:

Is kOS allowed?

kOS feels a bit unfair given that it allows easy, simple closed loop control and global keolocation and thus completely eliminates the innaccuracy problem, rejecting all the butterfly effect of mid-course disruptions caused by unpredictable bouncing off the terrain and race conditions in KSP's physics that should normally be causing hundreds or even thousands of meters of midcourse inaccuracy. I feel like if I were doing a kOS rover challenge I would want to make the target location randomized by a kOS script.

It would still be interesting to see how far you can make it with kOS but I think it's too big of an unfair advantage unless you can use it in a way that doesn't have access to information a KAL doesn't have access to.

 

Edited June 27, 2022 by Pds314

[18Watt]

Well, I got past the first mountain- the one that starts at about 06N.  We'll see how far I make it.

I think in order to get close to the pole I will need to figure out a way to do simple course changes based on time.  @Pds314's suggestion of mounting a control element (like docking port or probe core) to a servo seems like a good way to go.

Having never really used the KAL before, can you do really long time intervals?  The entire journey will likely take 2 days (12 hours), just a really rough guess.

I have a rover design that can maintain a somewhat constant speed over non-mountainous terrain, so I think a time-based approach might actually work, if I can get the KALs to handle extremely long time intervals.

Having fun with this challenge!

 

[Pds314]

12 minutes ago, 18Watt said:

Well, I got past the first mountain- the one that starts at about 06N.  We'll see how far I make it.

I think in order to get close to the pole I will need to figure out a way to do simple course changes based on time.  @Pds314's suggestion of mounting a control element (like docking port or probe core) to a servo seems like a good way to go.

Having never really used the KAL before, can you do really long time intervals?  The entire journey will likely take 2 days (12 hours), just a really rough guess.

I have a rover design that can maintain a somewhat constant speed over non-mountainous terrain, so I think a time-based approach might actually work, if I can get the KALs to handle extremely long time intervals.

Having fun with this challenge!

 

IIRC there is nothing stopping a KAL from handling long intervals.

And if they couldn't, you could just daisychain them off each other.

I have managed to get 29.9 degrees north with the tank before chaotically blasting the wheels off in a crash. I might post that run if my current run fails.

I've begun modifying the speed of the vehicle in areas that are either untested or very hilly or far into the run on because it's perfectly doable to go 40+ in some areas but in others high 20s is dangerous and I neither want to drive 25 the whole way or crash 3 hours into a run because I'm going too fast.

Edited June 27, 2022 by Pds314

[18Watt]

4 minutes ago, Pds314 said:

I have managed to get 29.9 degrees north

That's towards the north side of the second, large, mountainous area?  That's awesome!

[18Watt]

Well, I made it past 19N, still going.  I'm still calling this a test run- I don't think I'll make it all the way, and have found a few problems with my rover:

• I had nose cones on the nose and tail, to make it more aerodynamic for water excursions.  (Or hydronamic? More efficient in water.  You know what I mean.).  Traveling through the mountains, both nosecones blew up, they can't handle extremely steep gradient changes I guess.  The rest of the rover is still working.
• As this journey will take several days, I'm powering my rover with a bunch of RTGs, inside a cargo bay.  At about 19N, two of the RTGs started showing signs of overheating, orange bars.  Hmm, this could be a problem I will need to address.

I won't be offended if I managed to disqualify myself on this run, here's what I did-  Not 100% sure why two RTGs started to overheat, but I was curious if opening the cargo bay doors would solve the problem.  It did indeed solve the overheating problem.  However, I did manually open and then close the cargo bay during this run, to see if that would cool the RTGs off.

The more I think about it, I feel this run needs to be disqualified.  Not 100% sure, but I suspect a KAL could be programmed to open the cargo doors for a minute every 2 hours or so.  So I think I have a way around the RTG overheat issue.  I'm not willing to mount them on the exterior because there's a lot of them, and I can't afford to lose many of them.

[Pds314]

1 hour ago, 18Watt said:

Well, I made it past 19N, still going.  I'm still calling this a test run- I don't think I'll make it all the way, and have found a few problems with my rover:

• I had nose cones on the nose and tail, to make it more aerodynamic for water excursions.  (Or hydronamic? More efficient in water.  You know what I mean.).  Traveling through the mountains, both nosecones blew up, they can't handle extremely steep gradient changes I guess.  The rest of the rover is still working.
• As this journey will take several days, I'm powering my rover with a bunch of RTGs, inside a cargo bay.  At about 19N, two of the RTGs started showing signs of overheating, orange bars.  Hmm, this could be a problem I will need to address.

I won't be offended if I managed to disqualify myself on this run, here's what I did-  Not 100% sure why two RTGs started to overheat, but I was curious if opening the cargo bay doors would solve the problem.  It did indeed solve the overheating problem.  However, I did manually open and then close the cargo bay during this run, to see if that would cool the RTGs off.

The more I think about it, I feel this run needs to be disqualified.  Not 100% sure, but I suspect a KAL could be programmed to open the cargo doors for a minute every 2 hours or so.  So I think I have a way around the RTG overheat issue.  I'm not willing to mount them on the exterior because there's a lot of them, and I can't afford to lose many of them.

KAL can indeed open and close cargo bay doors as desired. But manually opening them in response to heat definitely isn't an automated run unfortunately.

[Pds314]

I created a course I thought would put the tank on relatively flat terrain to the pole, but accidentally sent the tank through some terrain more rugged than intended, resulting in it eventually failing.
The tank made it 33.396092 degrees north. It is pretty badly beat up, minus two Kerbals and with the third loose in the cabin, and with a lot of part connections yielded from the tumbling, and upside down.

Jumping 7 meters in the air. Tank is already missing a few things by this point, but I was AFK for that.


Perfect landing.


9 meters above the ground and falling fast.


Another excellent landing.


It trips on something and tumbles.


Loosing some parts, as well as stranding Jeb and Val, knocking Bill out of his seat, and crushing in the front right edge of the tank, including busting two wheels. That means the only thing piloting it is now the avionics nosecone and docking port. Fortunately the broken wheel yielded a little upward, which means it isn't dragging most of the time.


That is some serious damage.. The wheel periodically clips and causes the tank to do donuts on the low friction wheels.


Unfortunately, it flipped the tank, leaving Bill trapped in an immobile upside down busted tank quite far from the KSC. Task failed successfully!


Yeah... That might be a bit far to walk back from..

 

Edited June 28, 2022 by Pds314

[QF9E]

7 hours ago, Pds314 said:

kOS feels a bit unfair given that it allows easy, simple closed loop control and global keolocation and thus completely eliminates the innaccuracy problem, rejecting all the butterfly effect of mid-course disruptions caused by unpredictable bouncing off the terrain and race conditions in KSP's physics that should normally be causing hundreds or even thousands of meters of midcourse inaccuracy. I feel like if I were doing a kOS rover challenge I would want to make the target location randomized by a kOS script.

It would still be interesting to see how far you can make it with kOS but I think it's too big of an unfair advantage unless you can use it in a way that doesn't have access to information a KAL doesn't have access to.

I figured as much. At the same time, I'm having fun with kOS at the moment, so I will continue to use that. Yesterday I made an autothrottle that responds to the bumpiness of the terrain to prevent my rover from being lauched into the air. I had lots of fun just watching my rover scale mountains and survive.

I also have some ideas to let kOS handle the navigation on its own. My first attempt used manually selected waypoints, but I think there might be a way to do without them. We'll see how far I can get with it, even as a gatechrasher submission to this challenge.

Byt the way: I'm really impressed how far you and @18Watt have got with just the KAL!

[18Watt]

4 minutes ago, QF9E said:

I made an autothrottle that responds to the bumpiness of the terrain to prevent my rover from being lauched into the air.

That's exactly what I need!  Ok, I'll do it stock, but I can dream, right?

 

4 minutes ago, QF9E said:

Byt the way: I'm really impressed how far you and @18Watt have got with just the KAL!

Actually, I haven't used the KAL yet...  I'm just 'pointing north' and 'hoping for the best'.

 

Edited June 28, 2022 by 18Watt

[Pds314]

27 minutes ago, QF9E said:

I figured as much. At the same time, I'm having fun with kOS at the moment, so I will continue to use that. Yesterday I made an autothrottle that responds to the bumpiness of the terrain to prevent my rover from being lauched into the air. I had lots of fun just watching my rover scale mountains and survive.

I also have some ideas to let kOS handle the navigation on its own. My first attempt used manually selected waypoints, but I think there might be a way to do without them. We'll see how far I can get with it, even as a gatechrasher submission to this challenge.

Byt the way: I'm really impressed how far you and @18Watt have got with just the KAL!

I genuinely wonder if there's a way to make useful sensors to control a KAL. Like, maybe using docking ports that undock periodically, and if they're undocked for some period of time, a second KAL controlling the throttle will throttle down or something?

Or multiple crafts moving together with one targeting the other so that SAS manipulates a control surface in some particular way. I.E. If the vehicle is on a slope, it throws the control surface around and does something.

Edited June 28, 2022 by Pds314

[18Watt]

11 hours ago, Pds314 said:

I genuinely wonder if there's a way to make useful sensors to control a KAL.

The KAL would certainly have more uses if there were other inputs.  Right now I think time is the only input.

[Pds314]

5 hours ago, 18Watt said:

The KAL would certainly have more uses if there were other inputs.  Right now I think time is the only input.

Well or player action/axis groups. Or other KALs, but there's not really sensor inputs.

Edited June 28, 2022 by Pds314

[18Watt]

@Pds314, another question:

When I do my actual run, my original plan was to take screenshots in map view to show my approximate route.  However, about half of the run will occur at night, so that won't work very well.

Would it be OK to manually generate KerbNet waypoints as I go along?  The purpose would be so that after the voyage is complete, I can warp Kerbin so my route is on the day side, to give a rough idea of the route I took.

Yes, I think the KAL can generate Kerbnet Waypoints too, however that still requires manual intervention, you need to click ok to accept each generated waypoint.  Either way, I wouldn't be doing anything to affect the path of the rover, just trying to leave a trail for other players to see the route I took.

PS - No rush on this question- I'll be busy at work for the next few days, probably won't be able to do my actual run till this weekend..

 

Edited June 29, 2022 by 18Watt

[Pds314]

1 hour ago, 18Watt said:

@Pds314, another question:

When I do my actual run, my original plan was to take screenshots in map view to show my approximate route.  However, about half of the run will occur at night, so that won't work very well.

Would it be OK to manually generate KerbNet waypoints as I go along?  The purpose would be so that after the voyage is complete, I can warp Kerbin so my route is on the day side, to give a rough idea of the route I took.

Yes, I think the KAL can generate Kerbnet Waypoints too, however that still requires manual intervention, you need to click ok to accept each generated waypoint.  Either way, I wouldn't be doing anything to affect the path of the rover, just trying to leave a trail for other players to see the route I took.

PS - No rush on this question- I'll be busy at work for the next few days, probably won't be able to do my actual run till this weekend..

 

If it doesn't affect the run I would say any intervention for documentation purposes is fine.

[Pds314]

These hills are infuriating.

They have so many random little bumps that can flip or throw a vehicle that I've cut my prop angle (yes, my tank is prop driven because that's what SAS can control) down to 4 degrees.

Here's what happens every 3.8 hours right now:

 

I'm having to slow way down and even then it's no guarantee it won't crash.

 

Edited June 30, 2022 by Pds314

[18Watt]

19 hours ago, Pds314 said:

These hills are infuriating.

My plan right now is to follow the coast up before turning north.  I think I see a path that keeps me clear of severe terrain.

I'm cruising at an average speed of about 19.7 m/s, a little faster or slower on hills.  I'm not altering my prop settings at all.  Steering is via a probe core mounted on a servo, which is controlled by a KAL.  

20 m/s is painfully slow, but it greatly reduces the risk of tipping over.  Another advantage of avoiding severe terrain is the speed does not vary as much, which is important because every turn is time-based, from the moment I leave the runway.

The KAL is performing a few other functions- I need to occasionally open the cargo doors to cool off the RTGs.  I'm also pulling up KerbNet periodically to drop markers.  I still need to click to drop the marker, but at least the KAL will remind me to do that at regular intervals.

Oh, my rover is amphibious.  I thought I'd need to cross some rivers, but so far haven't seen any.  I am cutting across the bay north of the runway though.

Edit: @Pds314, are you sometimes experiencing the complete destruction of your rover for no apparent reason?  I'm occasionally getting that, but am never watching the screen when it happens.  The F3 screen says that every part collided with Kerbin and was destroyed.  At 20 m/s, I'd expect at least some of the parts to survive.  Not sure if I'm dropping into terrain gaps, hitting trees, or what.  I initially suspected the RTGs were overheating, but I've solved that problem.  Seems to be very random..

Edited July 1, 2022 by 18Watt

[Pds314]

4 hours ago, 18Watt said:

My plan right now is to follow the coast up before turning north.  I think I see a path that keeps me clear of severe terrain.

I'm cruising at an average speed of about 19.7 m/s, a little faster or slower on hills.  I'm not altering my prop settings at all.  Steering is via a probe core mounted on a servo, which is controlled by a KAL.  

20 m/s is painfully slow, but it greatly reduces the risk of tipping over.  Another advantage of avoiding severe terrain is the speed does not vary as much, which is important because every turn is time-based, from the moment I leave the runway.

The KAL is performing a few other functions- I need to occasionally open the cargo doors to cool off the RTGs.  I'm also pulling up KerbNet periodically to drop markers.  I still need to click to drop the marker, but at least the KAL will remind me to do that at regular intervals.

Oh, my rover is amphibious.  I thought I'd need to cross some rivers, but so far haven't seen any.  I am cutting across the bay north of the runway though.

Edit: @Pds314, are you sometimes experiencing the complete destruction of your rover for no apparent reason?  I'm occasionally getting that, but am never watching the screen when it happens.  The F3 screen says that every part collided with Kerbin and was destroyed.  At 20 m/s, I'd expect at least some of the parts to survive.  Not sure if I'm dropping into terrain gaps, hitting trees, or what.  I initially suspected the RTGs were overheating, but I've solved that problem.  Seems to be very random..

Nothing like that. I'm actually yet to lose any metal. Just lights, wheels, Kerbals, panels and control surfaces.

I have not had any confirmed collisions with trees yet. Three that were within less than 5 meters though.

 

Are all the collisions simultaneous?

Does it say you achieved some obscenely high altitude in F3?

Are any of your parts durable enough to survive top apeed impacts?

What happens if you drop-kick your rover off the top of the VAB? Does it jitter or explode or generate phantom forces?

Edited July 1, 2022 by Pds314

[18Watt]

7 hours ago, Pds314 said:

I have not had any confirmed collisions with trees yet. Three that were within less than 5 meters though.

I managed to hit one dead-center one time.  Otherwise, I'm amazed at how many I pass close to, but never hit.

7 hours ago, Pds314 said:

Are all the collisions simultaneous?

Basically yes.

7 hours ago, Pds314 said:

Does it say you achieved some obscenely high altitude in F3?

I didn't look, but will report the next time it happens.

7 hours ago, Pds314 said:

Are any of your parts durable enough to survive top apeed impacts?

The basic chassis is a MK2 cargo bay, 50 m/s impact rating.  Most of the delicate parts are inside the bay.  Additional tough parts are 2 nose cones (20 m/s?) and landing gear (really high impact rating).  I also have a couple of cubic octagonal struts, and several structural panels (which also have a crazy high impact rating).  

Most delicate parts are inside the bay- RTGs, probe core, batteries, reaction wheels, servo, and KALs.

7 hours ago, Pds314 said:

What happens if you drop-kick your rover off the top of the VAB? Does it jitter or explode or generate phantom forces?

LOL!  I'll try that, and report back.  Not sure how to drop-kick it..  Would just dropping it from a similar height accomplish the same test?  Thinking I could just F12 it to say 150 m altitude over the runway, after disabling 'ease into physics'.

Edited July 1, 2022 by 18Watt