I did not expect this to be as thoughtful as it is. Kudos to the thought put into the setup and training. Building custom hardware for the rat is amazing.. I need something like this for my own setup.
We reached the point of rat habituation but didn’t start training. Our rats (Todd, Kojima, Gabe) aged out before full testing. The setup works, but behavioral validation is pending.
>We reached the point of rat habituation but didn’t start training. Our rats (Todd, Kojima, Gabe) aged out before full testing. The setup works, but behavioral validation is pending.
Ah man, what a pity. That VR rig is awesome, but it doesn't really seem to me they are planning to continue these experiments, or do they?
Bad news: no video of them playing on this setup, just on the previous version. We iterated on v2 too long, our pet rats grew old and couldn't be trained. We open-sourced the hardware and software so others can build upon it. You can TLDR the whole thing in this thread: https://x.com/yolorun_capital/status/1996632980903620886?s=2...
...Also, here's my personal X, dm me if you have any questions, or would want to build it for a lab or for yourself: https://x.com/viktor_thoth
Doom aside (very impressive), I love the concept of putting rats on a roller ball instead of in a wheel for exercise. It would be better on their backs.
The year is 2034. Countless attempts at re-producing the sophisticated wetware of the brain have failed. Modeling research has proved unfruitful, with the curse of dimensionality afflicting every attempt at breaking the walls of general intelligence. With only a few million of capital left, and facing bankruptcy, they knew that only one option remained.
You really should release parts as parametric or at least the source files. I see everything is an STL and STLs are just a pain to work with. Suppose we want to try with mice? Or what about my cat? I do not expect just scaling in my slicer is going to end up with a good result, I'll need to redo everything from scratch. But parametric parts? That gives us a lot faster iteration. That gives you a lot faster iteration too! I highly recommend taking that approach when designing and I find it is worth it more often than not.
Could you add cost estimates to the BOM? These never need to be accurate but I always find it helpful when estimating a project. You're just saving people from the time it takes to click every single link and throw them into a calculator. And informs people very quickly what to innovate on to drive costs down. (Sorry, BOMs without cost estimates are a big pet peeve of mine)
# Questions:
- Do the rats enjoy playing Doom?
- Are there specific games the rats like to play?
I've never thought about what types of videogames other animals would enjoy, but damn if you didn't just open Pandora's Box here. I actually think we could learn a lot about them (and even their specific personalities) from this question. It gives a whole other level of refinement than just knowing what my cat's favorite toys and games are...
And also, thanks for open sourcing this! I'm excited to see what comes of it!
As an evolutionary cousin of a rat, the half-second delay after firing my weapon and the response would make me want to eat my whiskers.
This would completely kill any potential reward (and replace it with the opposite, frustration) you're trying to train me with, please fix immediately.
Whats the fascination with Doom? I keep seeing running Doom on this and that and now Rats playing Doom. I mean ok, Doom runs on everything and now rats playing doom. So what?
Doom has some star power and is instantly recognizable to pretty much anyone, not just video game fans. Its engine is Free software that's easily adaptable to pretty much any input or output method. In this case Doom might also have made a better candidate than another first person shooter because you can't aim up or down in Doom, so there's no need to account for that in the rats' input.
Wonderfully crazy. Should the rats have multiple weapons at their disposal, which do they prefer? How many demons escape their fate? Are there videos of gaming sessions? So many questions!
68 comments
[ 13.5 ms ] story [ 80.4 ms ] threadAh man, what a pity. That VR rig is awesome, but it doesn't really seem to me they are planning to continue these experiments, or do they?
Bad news: no video of them playing on this setup, just on the previous version. We iterated on v2 too long, our pet rats grew old and couldn't be trained. We open-sourced the hardware and software so others can build upon it. You can TLDR the whole thing in this thread: https://x.com/yolorun_capital/status/1996632980903620886?s=2...
...Also, here's my personal X, dm me if you have any questions, or would want to build it for a lab or for yourself: https://x.com/viktor_thoth
hehe very important
although... if the walls moved and touched the whiskers in conjunction with the game could be something
"Bring me the rats."
# Suggestion:
You really should release parts as parametric or at least the source files. I see everything is an STL and STLs are just a pain to work with. Suppose we want to try with mice? Or what about my cat? I do not expect just scaling in my slicer is going to end up with a good result, I'll need to redo everything from scratch. But parametric parts? That gives us a lot faster iteration. That gives you a lot faster iteration too! I highly recommend taking that approach when designing and I find it is worth it more often than not.
Could you add cost estimates to the BOM? These never need to be accurate but I always find it helpful when estimating a project. You're just saving people from the time it takes to click every single link and throw them into a calculator. And informs people very quickly what to innovate on to drive costs down. (Sorry, BOMs without cost estimates are a big pet peeve of mine)
# Questions:
- Do the rats enjoy playing Doom?
- Are there specific games the rats like to play?
I've never thought about what types of videogames other animals would enjoy, but damn if you didn't just open Pandora's Box here. I actually think we could learn a lot about them (and even their specific personalities) from this question. It gives a whole other level of refinement than just knowing what my cat's favorite toys and games are...
And also, thanks for open sourcing this! I'm excited to see what comes of it!
This would completely kill any potential reward (and replace it with the opposite, frustration) you're trying to train me with, please fix immediately.