Embodied Playground: HKU-led team launches long-horizon robotics benchmark with 410 environments
YiMaTweets · x · 2026-10-22
A team from HKU, Sharpa, Tsinghua, Stanford, MIT and UNC launches Embodied Playground, a long-horizon agentic robotics benchmark testing how foundation-model agents coordinate reasoning with physical execution when each action shapes the next decision.
- Scale: 410 task environments, 100 curated tasks, 5 evaluation suites, 16 task types
- Three axes: physical (contact/mass/friction/motion prediction), strategic (legal intermediate states, ordering dependent actions, revising plans), spatial (pose tracking, interlocking assembly, grasp/placement)
- Agents read the scene, write control code, and act through a robot under declared observation/tool constraints and execution budgets
- Example tasks: projectile chain reactions, legal Tower of Hanoi, Kongming lock assembly, cube solving via grasping/turning, multi-step household manipulation
EP-100 results are already published.
Related event: HKU-led Team Releases Embodied Playground Benchmark(2 posts)→
More from Embodied
- SenseTime's 12-benchmark study: GPT-6-Astra hits SOTA in sim but costs ~$55 to place pens — heyshrutimishra · 2026-10-22
- Tuesday Labs shows off Voxel, a robot butler starting out by cleaning crumbs — CurieuxExplorer · 2026-10-22
- GPT-6-Astra nears saturation on 12 robot benchmarks but burns 9M tokens per task — rohanpaul_ai · 2026-10-22
- Musk: Tesla will carry the majority of passenger miles on Earth wherever allowed — alexmacgregor__ · 2026-10-22
- Flying a drone with Codex: 12–20s per decision is the bottleneck — kfountou · 2026-10-22
- Robotics fashion is now an industry, observer says — micoolcho · 2026-10-22