The CoRL 2026 workshop on Compositional Generalization for Real-World Robot Learning explores how modular and compositional approaches can enhance robot learning by combining scaled systems with structured, reusable components. It focuses on challenges in modality integration, module interfacing, agentic systems, and benchmarking compositionality, targeting researchers in robotics, machine learning, and embodied AI.
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short papers and posters
Reviewed for technical correctness, reproducibility, and actionable artifacts (e.g., skill libraries, benchmarks, datasets, metrics, or well-documented system case studies); welcomes works-in-progress and negative/ablation results.
Research areas in scope
01
Core Challenges & Topics
Modality compositionInterfacing modulesThe role of agents in roboticsEvaluation and benchmarking for compositionality
Policies worth checking twice
Submissions are non-archival.
Work already published or accepted to the CoRL 2026 main conference will not be accepted.
Authors are encouraged to share code and evaluation details.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.