Footbots moving in 3D space

Requests regarding how to set up experiments in ARGoS.
jharwell
Posts: 67
Joined: Thu Jan 25, 2018 4:12 am

Footbots moving in 3D space

Postby jharwell » Fri Oct 11, 2019 5:00 pm

Hello,

I'm currently trying to figure out how to get the footbots to move in 3D space in ARGoS. From my reading of the code, it seems like this should fully work by simply using a 3D physics engine. What I'm currently stuck on is how to get the robots to drive UP onto something (such as a ramp). ARGoS does not seem to have native support for something like that, and it looks like my options are:

1. Add a ramp entity, basing it on the box entity, and then give it a 3D physics model. How difficult would this be to do? I can follow what was done on the box entity, but this is pretty far outside my area of expertise, and so I'm not sure how difficult it would be to get working in practice.

2. Use the prototype entity (which was originally designed for robot links and joints) to create a body using a convex hull (the option to do this exists in the code, but

Code: Select all

argos3 -q prototype
does not list the convex hull as one of the supported link types. Is this reliable, and would it be computationally more expensive that option #1 if I created hundreds or thousands of such prototypes in ARGoS.

Thanks!

pincy
Site Admin
Posts: 632
Joined: Thu Mar 08, 2012 8:04 pm
Location: Boston, MA
Contact:

Re: Footbots moving in 3D space

Postby pincy » Fri Oct 11, 2019 9:54 pm

The author of the 3D engine and of the prototype entity is Michael Allwright. I'll ask him to have a look at this question.
I made ARGoS.

mallwright
Posts: 35
Joined: Tue Apr 17, 2012 11:15 am

Re: Footbots moving in 3D space

Postby mallwright » Mon Oct 14, 2019 9:37 am

Sorry for the delay in replying. I had some issues with my account.

Convex hull support is provided via the prototype entity as a link geometry. I haven't added it as a standalone entity. Support for this geometry isn't going anywhere, I just forgot to update the documentation.

Regarding your requirements, does the ramp need to move in the environment? If not (i.e., it can be fixed), you can just use a thin rotated box entity as a ramp with the attribute movable set to false so that it does not collapse. The computational complexity of detecting collisions for a box shape is cheaper than detecting collisions for a convex hull, but it is hard to say whether this difference will be the bottleneck in your simulation.

Just a heads up, the foot-bot model currently does not have a 3D model and I wasn't planning on adding one since this hardware is almost completely out of date. I need to add a 3D model for the e-puck robot for our work, perhaps you could use that for your experiments?

pincy
Site Admin
Posts: 632
Joined: Thu Mar 08, 2012 8:04 pm
Location: Boston, MA
Contact:

Re: Footbots moving in 3D space

Postby pincy » Mon Oct 14, 2019 2:01 pm

The e-puck is an even older robot, with fewer sensors than the foot-bot. I think we should have a student make a 3D model of the foot-bot.
I made ARGoS.

mallwright
Posts: 35
Joined: Tue Apr 17, 2012 11:15 am

Re: Footbots moving in 3D space

Postby mallwright » Tue Oct 15, 2019 9:51 am

Do you have someone in mind? The E-puck may be older but it is still being used at our lab for research. The foot-bots are just collecting dust.

pincy
Site Admin
Posts: 632
Joined: Thu Mar 08, 2012 8:04 pm
Location: Boston, MA
Contact:

Re: Footbots moving in 3D space

Postby pincy » Tue Oct 15, 2019 5:23 pm

I don't think that's how we should prioritize.

The simulated foot-bot is used by hundreds of people and is the most feature rich robot we have. The fact that at IRIDIA you don't use it does not mean that we should discontinue it in its simulated form. Also, the e-puck is a purely tabletop robot - simulating it in 3D would translate to no practical use. It would be a good example on how to make a simple 3D model though, and if the 3D engine works a bit like the 2D one, lots of code would be in common between e-puck and foot-bot.

I don't have anybody in mind to do this because I didn't do the 3D engine. It would be great if it were someone close to you, since you know the insides of the code better than anyone. If not, I'll try to find a student to do it here. The problem usis that I have zero bandwidth for this kind of stuff because I'm about to start teaching.
I made ARGoS.

Nealrin
Posts: 1
Joined: Thu Oct 17, 2019 9:14 am

Re: Footbots moving in 3D space

Postby Nealrin » Sat Oct 19, 2019 12:56 pm

How old are your students, Pincy? Might be a great assignment for some of the more advanced of them.

pincy
Site Admin
Posts: 632
Joined: Thu Mar 08, 2012 8:04 pm
Location: Boston, MA
Contact:

Re: Footbots moving in 3D space

Postby pincy » Sat Oct 19, 2019 2:20 pm

19 to 29, age and experience are not a problem in theory. I just need to find the right student.
I made ARGoS.

jharwell
Posts: 67
Joined: Thu Jan 25, 2018 4:12 am

Re: Footbots moving in 3D space

Postby jharwell » Mon Mar 02, 2020 10:28 pm

Have you found a student to work on the foot-bot dynamics3d model? Just wondering if I should plan on adapting my 3D experiments with the foot-bot to work with the pointmass3d engine or not.

Thanks!
- John

pincy
Site Admin
Posts: 632
Joined: Thu Mar 08, 2012 8:04 pm
Location: Boston, MA
Contact:

Re: Footbots moving in 3D space

Postby pincy » Tue Mar 03, 2020 8:28 am

Not at the moment, I'm in the middle of teaching two courses. The summer is usually a good time for these projects.
I made ARGoS.


Return to “How to... ?”