Aidan

PhD @ Mila and McGill | Paradigms of Intelligence Team @ Google | Deep learning, Recursive Reasoning, Embodied Intelligence, Computational Neuroscience

[Aidan @aidan_sirbu]

1/14) Virtual animals are a long-standing goal in neuro-AI, but they should actually behave like real animals.

We present SCAMPER: a virtual rat for RL studies that moves naturally during both exploration, and goal-directed behavior.

Project page: https://t.co/s0J1igsgEv

[Blaise Agüera (@blaiseaguera.bsky.social) @blaiseaguera]

Proud to share new work from our Paradigms of Intelligence team toward AI designed for decentralized hardware, where low-power communication and fault tolerance are critical requirements. Using Neural Cellular Automata (NCA), @mayalen_etc and her co-authors show that spatial locality is not a barrier to complex multi-step reasoning.

Read the full paper on arXiv: https://t.co/B3M4E9CCdb

[Nick Blauch @nmblauch]

I've been spending the last several months stress testing this software in beta for my research. It's so incredibly feature rich and flexible for all sorts of different needs and applications.

Critically, it is super scalable but also amenable to academic grade hardware; the more compute you have the better, but the different backends can support fast sim training that is ready to convert to high fidelity training whenever more compute becomes available. That means one workflow, different presets.

This is a huge unlock for democratizing sim2real robotics research, but also - and this is extremely underappreciated - for theoretical neuroscience. If we want to understand the brain, we need to simulate how perception, cognition, and action come together in dynamic environments. That's what training robot policies in sim is all about, and no headaches about hardware.

Somehow, we've built superintelligent AI far before we've understood it. Now, we can use AI to help us understand ourselves, from prototyping the task environment, to testing different embodiments, to implementing different theories of neural computation and different forms of experience.

Give it a shot and let us know how we can help. Isaac Lab is developed completely open source on GitHub.

[Aidan @aidan_sirbu]

@jialiiiz @tyrell_turing Although we don’t show it in the paper, I did also train a policy network that receives varying target velocities as observations, so during inference you could gradually ramp up the target speed. So to answer your question, the transition between gaits was quite smooth indeed!

[Aidan @aidan_sirbu]

@jialiiiz @tyrell_turing The videos shown of the different gaits were taken from unique policy networks each trained to move at varying target speeds.

[Will Knight @willknight]

oh hell yes. this is EXACTLY what I want my agentic mega-swarm to look like.

[Charles Zhang @chrlszhng]

Proud to be part of this! SCAMPER builds on MIMIC-MJX with PULSE-like skill distillation (@zhengyiluo), improving latent motion control to enable naturalistic behavior for simulated neuroscience experiments.

[Blake Richards @tyrell_turing]

Check out our new model for creating realistic motion in a virtual rat during RL!

I think of this as providing the "brainstem and spinal cord" for NeuroAI researchers who want to simulate realistic neuroscience experiments.

Congrats to @aidan_sirbu for getting this out!

[Aidan @aidan_sirbu]

14/14) Check out the full paper: https://t.co/QOZKwNITBm

And come check out our work at NeurIPS in Sydney!

A very big thank you to the great team without which this work would not have been possible: @chrlszhng @Scottyang417 @talmop @BenceOlveczky @tyrell_turing

[Aidan @aidan_sirbu]

1/14) Virtual animals are a long-standing goal in neuro-AI, but they should actually behave like real animals.

We present SCAMPER: a virtual rat for RL studies that moves naturally during both exploration, and goal-directed behavior.

Project page: https://t.co/s0J1igsgEv

[Aidan @aidan_sirbu]

13/14) Natural motion exploration matters for neuroscience and SCAMPER provides the neuro-AI community a tool to achieve this in virtual animals. This brings us a step closer to simulating real animal experiments in silico.