RoboDojo benchmarks 30 robot manipulation policies across 42 sim and 18 real tasks
jiqizhixin · x · 2026-07-27
RoboDojo is a unified benchmark for evaluating robot manipulation policies in both simulation and the real world.
- Built by 18+ top labs, including MMLab@HKU, UC Berkeley, MIT, and Stanford.
- Covers 42 simulation tasks and 18 real-world tasks across 5 core skills: generalization, memory, precision, long-horizon execution, and open-vocab instructions.
- The authors tested 30 policies on a reproducible platform with cloud-based real-world access.
- They claim it improves on prior benchmarks in scale, reproducibility, and real-world fidelity.
The paper and project position RoboDojo as an open-source benchmark for comparing generalist robot manipulation systems more realistically than sim-only setups.
More from Embodied
- XPENG's IRON robot demos full-duplex speech with 9-mic array and lip reading — ChrisGPT · 2026-09-23
- Uber riders can now hail Waymo robotaxis on Austin freeways — ATTlKA · 2026-09-23
- M5 Ultra LLM test: 4x faster prompt processing, but double the power draw — DigitalguyCH · 2026-09-23
- Figure CEO: Humanoid Robotics Needs Four Stages and Eventually Hundreds of Billions — adcock_brett · 2026-09-23
- InstinctFlash: open-source runtime runs 8 robot model families in real time on one commercial GPU — chris_j_paxton · 2026-09-23
- M5 Ultra 96GB vs RTX 5090-6000 Pro: which rig for local video generation at $6.5K-$27K — MaxwellHusk · 2026-09-23