aiqbench analysis
ATOM-Bench introduces a critical diagnostic framework that separates failures in robotic manipulation into weak atomic skills versus limited compositional generalization, addressing the gap between demonstrated success and real-world adaptability. By factorizing tabletop tasks into motor and instruction atoms across single- and dual-arm systems, this benchmark provides essential data to evaluate whether current foundation models can truly recombine learned behaviors for novel structures. The release of 3,000 human demonstrations alongside physical rollout results offers a necessary testbed for advancing robust robotic control policies beyond simple task execution.
ATOM-Bench: A Real-World Benchmark for Atomic Skills and Compositional Generalization in Manipulation Policies
arXiv:2606.16826v1 Announce Type: cross Abstract: Generalist manipulation policies are increasingly presented as foundation models for robotic control, but their real-world generalization remains difficult to diagnose. A policy may succeed on demonstrated tasks while still failing to execute fine-grained atomic skills or recombine learned skills in new task structures. We introduce \textbf{ATOM-Bench}, a real-world benchmark for evaluating both atomic skills and compositional generalization in manipulation policies. ATOM-Bench factorizes tabletop manipulation into motor atoms and instruction atoms, and contains 30 atomic tasks and 24 held-out compositional tasks across paired single-arm and dual-arm robot tracks. We collect 3,000 human demonstrations for atomic fine-tuning and release both the demonstration data and evaluation rollout data to support reproducible real-world evaluation. Policies are fine-tuned on atomic tasks and evaluated on both atomic skill acquisition and held-out compositional tasks. We further introduce Atomic Score (AS) and Compositional Failure Share (CFS) to distinguish failures caused by weak atomic skills from failures caused by limited compositional reuse. Through 2,700 physical rollouts on five representative manipulation policies, we find that current policies can acquire simple instruction-grounding skills, but still struggle with fine-grained motor atoms, counting, and logical filtering. More importantly, strong atomic performance does not reliably transfer to held-out compositional tasks. ATOM-Bench provides a diagnostic testbed for studying whether failures arise from weak motor execution, poor instruction grounding, or limited compositional reuse.
Abstract:Generalist manipulation policies are increasingly presented as foundation models for robotic control, but their real-world generalization remains difficult to diagnose. A policy may succeed on demonstrated tasks while still failing to execute fine-grained atomic skills or recombine learned skills in new task structures. We introduce \textbf{ATOM-Bench}, a real-world benchmark for evaluating both atomic skills and compositional generalization in manipulation policies. ATOM-Bench factorizes tabletop manipulation into motor atoms and instruction atoms, and contains 30 atomic tasks and 24 held-out compositional tasks across paired single-arm and dual-arm robot tracks. We collect 3,000 human demonstrations for atomic fine-tuning and release both the demonstration data and evaluation rollout data to support reproducible real-world evaluation. Policies are fine-tuned on atomic tasks and evaluated on both atomic skill acquisition and held-out compositional tasks. We further introduce Atomic Score (AS) and Compositional Failure Share (CFS) to distinguish failures caused by weak atomic skills from failures caused by limited compositional reuse. Through 2,700 physical rollouts on five representative manipulation policies, we find that current policies can acquire simple instruction-grounding skills, but still struggle with fine-grained motor atoms, counting, and logical filtering. More importantly, strong atomic performance does not reliably transfer to held-out compositional tasks. ATOM-Bench provides a diagnostic testbed for studying whether failures arise from weak motor execution, poor instruction grounding, or limited compositional reuse.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2606.16826 [cs.RO]
(or arXiv:2606.16826v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.16826 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 313 words
Related coverage
LabDex: A Hierarchical Benchmark for Dexterous Manipulation in Laboratories
arXiv · Aug 20, 2026
[AINews] Claude Fable/Mythos 5.1: new SOTA model, 75% cache price cut but 70% more output tokens
Latent Space · Sep 2, 2026
SkillMemo: Expert-guided Skill Memory Framework for Compositional Embodied Manipulation
arXiv · Aug 7, 2026
MobilePA-Bench: Benchmarking Mobile Planner Agents on Complex Real-World Tasks
arXiv · Aug 25, 2026