Event
EDAR: An Action Representation Method Coupling Actions with Environmental Effects
1 reports · 1 sources
Yuecheng Xu et al. propose EDAR (Environment-Relevant Action Representation), a method that anchors robot action tokens simultaneously to executable control structures and expected visual consequences, enabling the learned action space to capture interaction semantics rather than merely command-level patterns. The authors report that experiments on simulation and real-world manipulation benchmarks show EDAR improves downstream policy learning, especially on long-horizon manipulation tasks.
Report timeline
- EDAR: Learning Environment-Dependent Action Representations for Robotic Manipulation
EDAR (Environment-Dependent Action Representation) anchors action tokens in both executable control structure and expected visual consequences. By coupling motor commands with their environment-conditioned effects, it enables the learned action space to capture interaction semantics rather than merely command-level patterns.
arXiv cs.RO ·