CoRL - Heungwoo/research GitHub Wiki

CoRL — Conference on Robot Learning

CoRL is the natural sibling venue to ICLR for VLA work — typically more focused on real-robot systems and end-to-end learning recipes.

Surveys hosted in this wiki

Year Location Survey
CoRL 2026 🆕 Austin, Nov 9–12, 2026 [VLA & Manipulation Survey](/Heungwoo/research/wiki/CoRL-2026-VLA-Manipulation-Survey) (preliminary — 687 accepted; official list pending)
[CoRL 2025](/Heungwoo/research/wiki/CoRL-2025) Seoul, Sept 27–30, 2025 [VLA & Manipulation Survey](/Heungwoo/research/wiki/CoRL-2025-VLA-Manipulation-Survey) (live)

Baseline / flagship papers at CoRL

Foundational CoRL 2024 papers referenced across the wiki

  • OpenVLA (CoRL 2024) — the open-source 7B autoregressive VLA (Prismatic VLM + discretized action tokens on Open X-Embodiment); the reference baseline most later VLAs report against.
  • ReKep (CoRL 2024) — training-free manipulation via relational keypoint constraints written by a VLM and solved as real-time optimization.

← Back to Home