Jump to content

KSP 2 robotic parts  

61 members have voted

  1. 1. Do you think there will be robotic parts in KSP 2?

  2. 2. Do you want robotic parts in KSP 2?



Recommended Posts

Do you think there be robotic parts in KSP 2? Will they be added in the base game or in a DLC? 

I really hope they add robotic parts in KSP 2, so I can make cranes, excavators, and other equipment in KSP 2. If they do add it, I want them to be procedural, where I can make a hydraulic cylinder really long, really short, wide, or skinny. Also, telescopic crane booms similar looking to Liebherr's TELEMATIK system, and NON-FLOPPY ROTATION SERVOS, SOMETHING VERY SIMILAR TO CRANE SLEW RINGS.

Also for VTOLs and similar things.

Edited by Ben J. Kerman
Link to comment
Share on other sites

What I'd really like to see is robotic arms that use inverse kinematics so you can just tell the end of the arm where to go.  Even if it meant they were prebuilt arms, and thus not as versatile as the single-joint robotic parts, it'd be very nice to have something simple for use when assembling stations.

Link to comment
Share on other sites

5 minutes ago, leopardenthusiast said:

What I'd really like to see is robotic arms that use inverse kinematics so you can just tell the end of the arm where to go.  Even if it meant they were prebuilt arms, and thus not as versatile as the single-joint robotic parts, it'd be very nice to have something simple for use when assembling stations.

+1 for IK

Link to comment
Share on other sites

8 minutes ago, leopardenthusiast said:

Even if it meant they were prebuilt arms

Would actually not be hard to make it work for user-built stuff. UI for setting up joints might get a little awkward, but something like Jacobian Transpose method is a breeze to write, works with arbitrary chains, and does alright with joint limits.

Link to comment
Share on other sites

It would be deeply unfortunate for KSP2 not to have robotic parts at launch. I would consider it incomplete without them. Human space program has them, and KSP1 has them, so KSP2 should have them too. In fact, I'm about 99.9% certain I've seen them on at least one of the videos. You can see what looks almost certainly like a multi-joint arm right next to one of the interplanetary ships at its construction dock. I don't remember the particular video, but if you find that scene you can see the arms. 

Link to comment
Share on other sites

21 minutes ago, TLTay said:

It would be deeply unfortunate for KSP2 not to have robotic parts at launch. I would consider it incomplete without them. Human space program has them, and KSP1 has them, so KSP2 should have them too. In fact, I'm about 99.9% certain I've seen them on at least one of the videos. You can see what looks almost certainly like a multi-joint arm right next to one of the interplanetary ships at its construction dock. I don't remember the particular video, but if you find that scene you can see the arms. 

You're right, it's at 1:50 in the Gamescom 2019 gameplay video.  Not the greatest quality, but that is definitely some sort of robotic arm:

Spoiler

wzjFmIs.png

 

Link to comment
Share on other sites

3 minutes ago, TLTay said:

Find the connection between the ship and station. On the station side of that connection, look directly up. It's light grey.

HA

I thought that was pipes coming off the pillar in the background... thanks.

Link to comment
Share on other sites

8 hours ago, mazer924 said:

I don't know video was that but I can swear that there was a scene where there was some sort of arm (on a station) and it was moving away from a ship. So robotics is definitely there.

That's a pretty big leap. This comes from old footage - pre Intercept. The game has been pretty much re-written since then. There is no guarantee that this arm made it through to the current version. And I'm pretty sure we never have actually seen it move. Regardless, though, it could easily have very limited motion, similar to how we've had ramps before Breaking Ground, but you'd never consider it robotics.

At best, we can say that at some point during game's development, components that would make sense with robotics have been considered. This isn't anywhere near "definitely there," as much as I'd like to hope so.

Link to comment
Share on other sites

I have to agree with @K^2 Just because it's show in an early pre-alpha trailer doesn't mean that something is going to be in the game immediately apon release, if ever. (It would be cool if robotics are in the game at release. I'm personally not expecting it.) 

Link to comment
Share on other sites

I put my money on another "robotics DLC", which would be a win for almost everyone involved:

- Gives devs more time to focus on other aspects of the game at launch so core gameplay is ironed out

- Gives the devs/publisher a way to get other money to sustain the game after launch

- Gives devs more time to build the feature, with possible improvements from the previous iteration

- Keeps the game from getting delayed again due to less features required for launch

Those that lose out would be the fans that are too cheap to pay for a stock robotics DLC and are stuck with mod implementations, which is a sensible consolation in today's day and age, where building a highly anticipated game that gets delayed years isn't exactly cheap. 

 

I played KSP for years without robotics, and still use them very sparingly due to their complexity and my lack of patience. I can live without them being put into the game at launch if it means getting the game on time and optionally paying for them later. 

Link to comment
Share on other sites

On 6/27/2021 at 1:34 PM, K^2 said:

Would actually not be hard to make it work for user-built stuff. UI for setting up joints might get a little awkward, but something like Jacobian Transpose method is a breeze to write, works with arbitrary chains, and does alright with joint limits.

I did some investigation into writing a least-squares solver for IK. As a surveyor, I realized that it's functionally the same as minimizing the error in a survey traverse with fixed starting and ending points. So, I modeled a robotic arm in my domain-specific software (STAR*NET). As it turns out, the software refuses to solve anything when you set the error range to +/- 90 deg instead of a few arcseconds. So, no proof of concept there. Not having any further expertise, I gave up.

As it turns out, someone came up with the same idea a couple decades before I did. IIRC, it was the Buss & Kim paper referenced in this paper: http://www.math.ucsd.edu/~sbuss/ResearchWeb/ikmethods/iksurvey.pdf (Introduction to Inverse Kinematics with Jacobian Transpose, Pseudoinverse and Damped Least Squares methods)

Link to comment
Share on other sites

The robotic arms could just be for show, since the BAE is how you put things together. BAE would be much less of a pain to use than grappling arms, and you could attach your part in half a second instead of 15 minutes. Maybe the arms and their reach are how the limits of the orbital launch facility are determined. I think for many cases, this approach would be superior to actually manipulating parts into place by fiddling with joint interfaces. 

Link to comment
Share on other sites

58 minutes ago, FleshJeb said:

As it turns out, someone came up with the same idea a couple decades before I did. IIRC, it was the Buss & Kim paper referenced in this paper: http://www.math.ucsd.edu/~sbuss/ResearchWeb/ikmethods/iksurvey.pdf (Introduction to Inverse Kinematics with Jacobian Transpose, Pseudoinverse and Damped Least Squares methods)

All good stuff. I like JT specifically of because how trivial it becomes in quaternions. You literally just take cross-product between the vectors to end effector and to target from given joint, multiply by a rate, and constrain to the joint limits. Apply iteratively and that's it. The only downside is that it's unstable with respect to starting conditions, so if you have to IK from animation on every frame, it doesn't work great. But it's fantastic for robotics.

If you want stable solutions, it works best if you add energy cost to individual joints, and then do damped least squares, yeah! So long as your energy costs are sensible, small variations in starting positions will result in small variations in motion to target, which is great for animation work.

Link to comment
Share on other sites

10 minutes ago, K^2 said:

All good stuff. I like JT specifically of because how trivial it becomes in quaternions. You literally just take cross-product between the vectors to end effector and to target from given joint, multiply by a rate, and constrain to the joint limits.

This applies strictly to rotations, yes? One of the reasons I liked the least squares solution is that it also handles telescoping segments. What I left out of my initial post is that I also measure distances, so that provides another error to play with. Again, the error my software is able to handle is rather small (typical value is 2mm + 2ppm for an infrared laser).

Link to comment
Share on other sites

2 hours ago, FleshJeb said:

This applies strictly to rotations, yes? One of the reasons I liked the least squares solution is that it also handles telescoping segments.

JT works with telescoping as well. If you do the math, the relevant component is just a dot product between error and the direction the piston extends in. Again, you want to scale it to relevant velocity and then apply limits to avoid over-extension. But it works pretty well.

JT actually works with arbitrary degrees of freedom. So long as you can take first derivative of your effector position with respect to your available DoF you can apply the method. For some DoF the math ends up very ugly, though. It's specifically translations and quaternion rotations that give you very clean, very easy to use results. Fortunately, that's also what you are usually dealing with in simple robotics and animation. If you have complex linkages, then, of course, you'll have to do additional work. But that tends to be the case with any IK method.

Link to comment
Share on other sites

×
×
  • Create New...