Simate's AutoResearch tops RoboDojo manipulation benchmark
Physical AI startup Simate took the number-one spot on the standardized RoboDojo manipulation leaderboard using AutoResearch, an AI-driven autonomous robotics research platform. By automating hypothesis formulation, code changes, and hardware evaluations through physical recursive self-improvement (Physical RSI), the platform demonstrates that agentic experimentation can outperform traditional human-engineered robot training pipelines.
Automated recursive self-improvement moving from software code generation into physical hardware execution marks a decisive shift in how embodied AI will scale.
- –Automating the hypothesis-to-hardware evaluation cycle bypasses the severe human trial-and-error bottleneck in robotics.
- –Ranking first on RoboDojo proves that autonomous agentic research can generalize across complex multi-task physical manipulation domains.
- –The competitive moat in robotics is rapidly shifting from human teleoperation volume to the speed and efficiency of automated physical experimentation engines.
DISCOVERED
1h ago
2026-09-25
PUBLISHED
1h ago
2026-09-25
RELEVANCE
AUTHOR
AI Revolution