mlsanityData to 7 Oct 2026

A Compositional Paradigm for Foundation Models: Towards Smarter Robotic Agents

arXiv 2510.18608 ↗First posted 21 Oct 2025cs.RO cs.LGcited by 0

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

The birth of Foundation Models brought unprecedented results in a wide range of tasks, from language to vision, to robotic control. These models are able to process huge quantities of data, and can extract and develop rich representations, which can be employed across different domains and modalities. However, they still have issues in adapting to dynamic, real-world scenarios without retraining the entire model from scratch. In this work, we propose the application of Continual Learning and Compositionality principles to foster the development of more flexible, efficient and smart AI solutions.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 14 of its 16 references that are in the map. First posted on arXiv 21 Oct 2025.

Builds on

14 of 16 references.

References · 14

  1. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 2,993
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  3. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,007
  4. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,653
  5. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,573
  6. 2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,168
  7. 2021CLIPort: What and Where Pathways for Robotic Manipulationcited by 617
  8. 2023Learning Interactive Real-World Simulatorscited by 440
  9. 2024ManiSkill3: GPU Parallelized Robotics Simulation and Rendering for Generalizable Embodied AIcited by 216
  10. 2022Instruction-driven history-aware policies for robotic manipulationscited by 93
  11. 2025The Future of Continual Learning in the Era of Foundation Models: Three Key Directionscited by 16
  12. 2024Has LLM Reached the Scaling Ceiling Yet? Unified Insights into LLM Regularities and Constraintscited by 3
2 more
  1. 2025Combining Pre-Trained Models for Enhanced Feature Representation in Reinforcement Learningcited by 2
  2. 2025GLAM: Efficient Continual Learning at Scale via Grouped LoRA Adapter Mergingcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Foundation Model Driven Robotics: A Comprehensive Review
  2. 2026Benchmarking World Models for Continual Learning on Compositional Tasks
  3. 2026Foundation Models in Robotics: A Comprehensive Review of Methods, Models, Datasets, Challenges and Future Research Directions
  4. 2026The Sim-to-Real Gap of Foundation Model Agents: A Unified MDP Perspective
  5. 2025Iterative Compositional Data Generation for Robot Control
  6. 2025Task-Agnostic Experts Composition for Continual Learning
  7. 2024AutoRT: Embodied Foundation Models for Large Scale Orchestration of Robotic Agents
  8. 2026Foundation World Models for Agents that Learn, Verify, and Adapt Reliably Beyond Static Environments