Lukun He

Research Scientist @AetherLab_AI | M.S. @ UC San Diego
Embodied intelligence, robot agents & robot foundation models.

[Lukun He @HeLukun11667]

🤖 Robots are getting better at manipulation. But how far are we from robot assistants that can operate autonomously in the open world?

We introduce RobotAgentBench (RAB), a benchmark for open-world robot autonomy in realistic household environments.

We evaluate several VLA/WAM models and diverse robot agent harnesses across five capabilities: task discovery, context reasoning, manipulation, failure recovery, and robustness. Four open-ended household scenarios test how these capabilities work together to fulfill high-level user needs.

🔍 What did we find?

• Manipulation ≠ autonomy. Agent harnesses improve performance, but reliable open-world autonomy remains challenging.

• Hybrid execution works best in our evaluation. Combining learned policies with code-based control achieves the strongest overall performance.

• Robot agents improve through recursive self-improvement. By refining reusable tools and accumulating experience, agents improve across iterations.

We’re excited to share this work and look forward to the community’s feedback!

🌐 Project, code & data: https://t.co/KVDXyrNfJs

📄 Paper: https://t.co/kFdRukZGsW

[Lukun He @HeLukun11667]

Excited to finally share CRIS-0, Aether AI’s first causal robotic intelligence system! 🤖

It’s been an incredible journey building this alongside such a talented team at Aether AI. Grateful to everyone who made this possible.

Still a long way to go, and excited for what’s ahead

[Aether AI (Causal Intelligence) @AetherLab_AI]

This is CRIS-0, Aether AI's first causal robotic intelligence system, operating in a real home. It reasons about how its actions will change the world, checks what actually happened, and adapts when reality doesn't go as expected.

[Lukun He @HeLukun11667]

@NanHuang99 wow, so cool!🎉