mlsanityData to 7 Oct 2026

Seven Sources of Physical AI Capability Formation

arXiv 2609.09627 ↗First posted 9 Sep 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Capabilities relevant to Physical AI can arise from materially different formation histories, yet existing taxonomies organized by morphology, architecture, learning algorithm, task, or domain do not directly answer what gives rise to a capability. We define a capability-formation source as a factor materially contributing to capability formation, distinct from components or construction steps. We identify seven non-exclusive sources: Recorded-Experience (RE), Predictive-Modeling (PM), Evaluative-Interaction (EI), Surrogate-Environment (SE), Mechanism-Grounded (MG), Embodied-Coupling (EC), and Evolution-Driven (ED) Formation. Using reconstructive induction with theoretical saturation, we traced a research matrix to primary studies, deduplicated the literature, set coding rules, and conducted three rounds of maximum-difference and negative-case sampling. Challenges included curriculum and self-supervised learning, active inference, open-ended and developmental learning, planning and search, neuro-symbolic architectures, digital twins, generative physical world models, and morphology-control co-design. Within the scope and criteria fixed as of September 4, 2026, all 49 evidence records were explainable by the seven sources individually or in combination. No R1-R3 challenge produced an irreducible eighth source, and R3 required no new core definition or substantive boundary rule. We therefore claim theoretical saturation within the stated scope, not logical completeness or exhaustive future coverage. The framework distinguishes similarity in observed capability from similarity in how it was formed, supporting analysis of explanation, transfer, replication, dependencies, governance evidence, and geoeconomic foundations.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 32 of its 47 references that are in the map. First posted on arXiv 9 Sep 2026.

Builds on

32 of 47 references.

References · 32

  1. 2017Deep reinforcement learning from human preferencescited by 3,846
  2. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 2,993
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  4. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,730
  5. 2022Do As I Can, Not As I Say: Grounding Language in Robotic Affordancescited by 2,302
  6. 2023PaLM-E: An Embodied Multimodal Language Modelcited by 2,037
  7. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,490
  8. 2018World Modelscited by 1,444
  9. 2024Octo: An Open-Source Generalist Robot Policycited by 1,427
  10. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,401
  11. 2018Learning Latent Dynamics for Planning from Pixelscited by 1,221
  12. 2018QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulationcited by 730
20 more
  1. 2019Hamiltonian Neural Networkscited by 702
  2. 2018Learning Dexterous In-Hand Manipulationcited by 643
  3. 2024Revisiting Feature Prediction for Learning Visual Representations from Videocited by 490
  4. 2023TD-MPC2: Scalable, Robust World Models for Continuous Controlcited by 482
  5. 2015Illuminating search spaces by mapping elitescited by 406
  6. 2019Learning agile and dynamic motor skills for legged robotscited by 402
  7. 2021Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learningcited by 395
  8. 2022DayDreamer: World Models for Physical Robot Learningcited by 372
  9. 2020Curriculum Learning for Reinforcement Learning Domains: A Framework and Surveycited by 350
  10. 2019End-to-End Safe Reinforcement Learning through Barrier Functions for Safety-Critical Continuous Control Taskscited by 256
  11. 2019Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutionscited by 182
  12. 2022Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behaviorcited by 113
  13. 2023RoboCat: A Self-Improving Generalist Agent for Robotic Manipulationcited by 108
  14. 2017Information Theoretic Model Predictive Control: Theory and Applications to Autonomous Drivingcited by 81
  15. 2017Scalable Co-Optimization of Morphology and Control in Embodied Machinescited by 38
  16. 2024Humanoid-Gym: Reinforcement Learning for Humanoid Robot with Zero-Shot Sim2Real Transfercited by 35
  17. 2023A Survey on Physics Informed Reinforcement Learning: Review and Open Problemscited by 22
  18. 2023Physics-informed Neural Networks to Model and Control Robots: a Theoretical and Experimental Investigationcited by 16
  19. 2017Embodied Evolution in Collective Robotics: A Reviewcited by 15
  20. 2020Environmental Adaptation of Robot Morphology and Control through Real-world Evolutioncited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026World Models for Embodied Intelligence: From Plausible to Controllable to Actionable
  2. 2026Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond
  3. 2026ACE-Brain-0.5: A Unified Embodied Foundational Model for Physical Agentic AI
  4. 2026From Perception to Action: Spatial AI Agents and World Models
  5. 2026From World Action Models to Embodied Brains: A Roadmap for Open-World Physical Intelligence
  6. 2026Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models
  7. 2025Aligning Perception, Reasoning, Modeling and Interaction: A Survey on Physical AI
  8. 2026The Institutional Scaling Law: Non-Monotonic Fitness, Capability-Trust Divergence, and Symbiogenetic Scaling in Generative AI