AM-Bench: A Simulation Suite and Benchmark for Aerial Manipulation
GuanyaShi · x · 2026-09-19
- The team released AM-Bench, a simulation suite and benchmark for aerial manipulation, submitted to CoRL 2026.
- Aerial manipulation is harder than ground manipulation: moving the arm shifts the drone, and wind pushes it off course, so policy and controller must cooperate.
- Key argument: for open-loop unstable systems like aerial manipulators and humanoids, policy learning is a system-level effort — action representation, high-level policy, low-level controller, and their interaction all matter.
- The benchmark aims to offer reusable insights into how these components fit together.
More from Embodied
- AMB3R-SLAM: Kilometer-Scale Monocular SLAM on One Consumer GPU, 70% Lower ATE — zhenjun_zhao · 2026-09-19
- EliGSiR: continual RGB-D Gaussian Splatting mapping under bounded compute hits 21.52 dB on TUM RGB-D — zhenjun_zhao · 2026-09-19
- SLAMSqueezeBench benchmarks nine SLAM systems under realistic edge compute and memory constraints — zhenjun_zhao · 2026-09-19
- ABC-130K: largest open bimanual teleop dataset with 3,500 hours, headed to CoRL 2026 — pabbeel · 2026-09-19
- 1X CEO interview: plans to ship 50k humanoids in 2027, world models and teleoperation — Scobleizer · 2026-09-19
- Snap Specs ($2,195) launches into Meta Ray-Ban backlash head-on against Apple's $1,999 folding iPhone — adariostrange · 2026-09-19