---
id: TECH-ROBOT-001
title: "Robotics and Embodied AI"
alternative_names:
  - "Robot Learning"
  - "Embodied Intelligence"
  - "Physical AI"
category: "Technical → Embodied Systems"
volume: III
chapter: 20
entry_type: "Technical Article"
level: 3
status: "Draft"
last_reviewed: 2026-10-15
version: "0.1.0"
prerequisites:
  - "TECH-RL-001"
related_entries:
  - "TECH-AGENT-001"
  - "TECH-LLM-001"
  - "TECH-DIFF-001"
  - "TECH-MM-001"
  - "TECH-DL-001"
  - "APP-PHYS-001"
  - "AGI-WM-001"
---

# Robotics and Embodied AI

## 1. Concise Definition
**Embodied AI** is the study and engineering of artificial agents that perceive and act through a physical (or physically simulated) body, so that their learning and behaviour are shaped by sensing, actuation and real-world dynamics. **Robot learning** is the subset that uses machine learning, as opposed to hand-engineered models and controllers, to produce robot behaviour.

## 2. Overview & Significance
Language and image models operate on digital data in which errors are cheap and data are plentiful. Robots must act in continuous, partially observed environments with contact dynamics, latency and safety constraints, and each real-world trial costs time and risks hardware damage. Embodied AI is therefore a distinct engineering regime, and it is a frequent test of claims about general intelligence, because physical competence requires integrating perception, planning and control (see [[AGI-WM-001] World Models](../vol05_agi/AGI-WM-001_world_models.md) and [[APP-PHYS-001] AI in Industry, Robotics and Autonomous Vehicles](../vol07_applications/APP-PHYS-001_ai_in_industry_robotics_and_autonomous_vehicles.md)).

Current research tries to import the recipe of large-scale pretraining from language and vision into robotics, producing general-purpose "vision-language-action" models (Brohan et al., 2023b; Kim et al., 2024). Whether this recipe will yield reliable general-purpose manipulation is open, and this article separates demonstrated results from claims.

## 3. Historical Development
Classical robotics decomposed the problem into state estimation, planning and feedback control built on analytic models. Machine learning entered first through perception and then through control. Levine et al. (2016) trained policies mapping camera images directly to motor torques, arguing for joint (end-to-end) training of perception and control. Reinforcement learning in simulation followed, with the principal obstacle being the *reality gap* between simulated and real dynamics. Tobin et al. (2017) proposed domain randomization, training in simulation with randomised rendering so that real images appear as another variation. OpenAI et al. (2019) used automatic domain randomization to train, in simulation only, a controller for a robot hand that solved a Rubik's cube. Lee et al. (2020) reported a locomotion policy trained in simulation and deployed on a legged robot over challenging terrain. Kaufmann et al. (2023) described Swift, a drone-racing system that combined simulation-trained reinforcement learning with real-world data and that, according to the authors, won several head-to-head races against three human champions and set the fastest recorded race time.

From 2022, large pretrained models were integrated. Ahn et al. (2022) (SayCan) used a language model to propose actions and learned value functions to ground them in what the robot could actually do. Driess et al. (2023) trained PaLM-E, an embodied multimodal language model. Brohan et al. (2023a) trained RT-1 on large-scale real robot data, and Brohan et al. (2023b) RT-2 by co-fine-tuning vision-language models on robot trajectories and web data. Open X-Embodiment pooled data across institutions (Open X-Embodiment Collaboration, 2024), and later open and commercial models followed (Kim et al., 2024; Black et al., 2024).

## 4. Technical / Conceptual Explanation

### 4.1 The control problem
A robot interacts with an environment as a partially observable Markov decision process (see [[TECH-RL-001] Reinforcement Learning](./TECH-RL-001_reinforcement_learning.md)): at each step it receives observations $o_t$ (camera images, joint states, force readings) and issues an action $a_t$ (joint targets, end-effector velocities, torques). A **policy** $\pi_\theta(a_t \mid o_{\le t}, \ell)$ may be conditioned on a language instruction $\ell$. Typical control rates are tens of hertz for manipulation and higher for dynamic locomotion or flight; large models must meet these latency budgets, often by predicting short *chunks* of future actions.

### 4.2 Learning paradigms
**Imitation learning (behaviour cloning)** fits $\pi_\theta$ to demonstration trajectories by supervised loss $\mathbb{E}\big[\|\,a_t - \pi_\theta(o_t)\,\|^2\big]$ or a likelihood loss. Errors compound because the policy drifts into states absent from the demonstrations. Zhao et al. (2023) addressed this for fine-grained bimanual tasks by predicting action sequences with a transformer (Action Chunking with Transformers) on low-cost hardware; the authors report learning six real-world tasks from about ten minutes of demonstrations and success rates of 80–90% on tasks such as slotting a battery. Chi et al. (2023) model the policy as a conditional denoising diffusion process, which can represent multimodal action distributions, and report an average 46.9% improvement over prior methods across 12 tasks from four benchmarks (see [[TECH-DIFF-001] Diffusion Models](./TECH-DIFF-001_diffusion_models.md)).

**Reinforcement learning** optimises expected return through trial and error, usually in simulation because real-world sample cost is prohibitive. **Sim-to-real transfer** relies on randomisation of physical and visual parameters (Tobin et al., 2017; OpenAI et al., 2019) or on adaptation methods surveyed by Zhao et al. (2020).

### 4.3 Foundation models in the loop
Two integration patterns exist. In **modular grounding** a language model produces a high-level plan and a library of learned skills and value functions filters it for feasibility (Ahn et al., 2022). In **end-to-end vision-language-action (VLA) models**, a pretrained vision-language model is trained to output robot actions represented as tokens or via a dedicated action head; RT-2 co-fine-tunes on robot trajectories and web-scale vision-language tasks, with actions written as text tokens (Brohan et al., 2023b). OpenVLA is a 7-billion-parameter open VLA trained on 970,000 real-world demonstrations; its authors report 16.5% higher absolute task success than the 55-billion-parameter RT-2-X across 29 tasks (Kim et al., 2024). The $\pi_0$ model combines a pretrained vision-language model with a flow-matching action head, trained on data from multiple robot types (Black et al., 2024).

### 4.4 Data
The principal bottleneck is robot data. Open X-Embodiment assembled data from 22 robots across 21 institutions, covering 527 skills and 160,266 tasks, and reported "positive transfer": a high-capacity model trained on pooled data improved performance on multiple robots relative to training on single-robot data (Open X-Embodiment Collaboration, 2024). Teleoperation, simulation and human video are alternative data sources.

## 5. Types, Components & Classifications
- **By embodiment**: fixed-base manipulators, mobile manipulators, bimanual systems, legged robots, humanoids, drones, autonomous vehicles.
- **By policy class**: regression-based, transformer-based, diffusion/flow-based, hybrid language-conditioned VLAs.
- **By learning signal**: imitation, RL, mixed pipelines combining pretraining, supervised fine-tuning and RL.
- **By integration**: modular (language planner plus skills) versus end-to-end.
- **By training venue**: real-world only, simulation only with transfer, or sim-plus-real.

## 6. Applications & Concrete Examples
**Deployed (outside the scope of learned generalist policies).** Industrial manipulators in structured cells and warehouse systems are widely used; most rely on conventional programming with learned perception components. This article does not catalogue such deployments.

**Research demonstrations with published evidence.** Legged locomotion over rough terrain (Lee et al., 2020); in-hand manipulation (OpenAI et al., 2019); drone racing against champions (Kaufmann et al., 2023); mobile-manipulator instruction following (Ahn et al., 2022); multi-task manipulation and instruction generalisation on real robots (Brohan et al., 2023a, 2023b); laundry folding, table cleaning and box assembly (Black et al., 2024, as reported by the authors).

**Not established.** Reliable, unsupervised household or open-ended general-purpose manipulation across arbitrary environments has not been demonstrated in the cited literature. Statements by developers about commercial humanoid or general-purpose robots are classified as claims pending independent evaluation.

## 7. Advantages, Limitations & Unresolved Questions
**Advantages.** Learned policies can handle perception noise and contact dynamics that are hard to model analytically, and pretrained models supply semantic generalisation, such as following instructions that do not appear in robot data (Brohan et al., 2023b).

**Limitations.** (1) Data scarcity and cost. (2) Reality gap in simulation-trained systems (Zhao et al., 2020). (3) Evaluation: results are reported on different robots, tasks and success criteria, with few shared physical benchmarks, so cross-paper comparisons are weak; the RT-2 and OpenVLA numbers above are developer-reported on their own task suites. (4) Safety: learned policies lack formal guarantees, and failures are physical. (5) Latency and compute: large models strain onboard control loops. (6) Hardware reliability and cost.

**Unresolved.** Whether pooled cross-embodiment pretraining scales as language pretraining did; the role of explicit world models versus end-to-end policies; how to certify safe behaviour for learned controllers; the relative value of simulation, teleoperation and human video as data sources.

## 8. Current State of Knowledge
As of 2026-10-15, the following are documented in peer-reviewed or archival sources: simulation-trained policies can transfer to real robots in specific tasks (Section 6), pretrained vision-language models can be adapted into robot policies and can improve generalisation to novel instructions in the authors' evaluations, and pooled multi-robot data can improve per-robot performance in at least the tested settings (see section 12). Independent, standardised evaluation of generalist robot policies is limited, and developer-reported success rates should be treated as Preliminary. The identities, specifications and prices of current commercial robot platforms change quickly and are not listed here; use `MODEL-001`, `FACT-001` and `TIME-001`.

## 9. Future Trajectory & Strategic Outlook
**Observed trends.** Policies have moved from single-task controllers to multi-task language-conditioned models trained on pooled data; diffusion and flow-matching action heads have become common.

**Forecast.**

| Field | Content |
| :--- | :--- |
| **Forecast ID** | `FCST-ROBOT-001` |
| **Forecast Statement** | A single learned policy will complete a diverse set of household-style manipulation tasks in homes it has not previously seen, at a success rate comparable to a trained human, as measured by an independent evaluator. |
| **Target Event** | Independent evaluation of a generalist manipulation policy in unseen real homes |
| **Operational Definition** | At least 20 tasks (e.g., loading a dishwasher, folding laundry, tidying) in at least 20 previously unseen homes, evaluated by an organisation independent of the developer, with mean per-task success at or above 90% and no unsafe events requiring intervention, using a published protocol. |
| **Time Horizon** | 2029–2033 |
| **Assumptions** | Continued growth of real-robot data; maturity of dexterous low-cost hardware; no sustained safety-driven moratorium. |
| **Confidence Level** | Low to Medium (roughly 25-40% by 2033) |
| **Alternative Scenarios** | Progress remains confined to structured settings (factories, warehouses); or reliability plateaus below deployment thresholds because of long-tail failures. |
| **Review Schedule** | Annual each October; final evaluation December 2033 |
| **Status** | `Active` |

**Medium and long term.** Beyond this record the article asserts no forecast.

## 10. Related Entries & Knowledge Graph
- **Foundational Concepts**:
  - [[TECH-RL-001] Reinforcement Learning](./TECH-RL-001_reinforcement_learning.md): The control-learning framework.
  - [[TECH-DL-001] Deep Learning](./TECH-DL-001_deep_learning.md): Perception and policy networks.
- **Related Methods & Technologies**:
  - [[TECH-AGENT-001] Autonomous AI Agents](./TECH-AGENT-001_autonomous_agents.md): Software agents sharing the perceive-plan-act loop.
  - [[TECH-LLM-001] Large Language Models](./TECH-LLM-001_large_language_models.md): Language backbones for planning and VLAs.
  - [[TECH-DIFF-001] Diffusion Models](./TECH-DIFF-001_diffusion_models.md): Basis of diffusion policies.
  - [[TECH-MM-001] Multimodal Foundation Models](./TECH-MM-001_multimodal_foundation_models.md): Vision-language pretraining.
- **Applications and Implications**:
  - [[APP-PHYS-001] AI in Industry, Robotics and Autonomous Vehicles](../vol07_applications/APP-PHYS-001_ai_in_industry_robotics_and_autonomous_vehicles.md): Deployment context.
  - [[AGI-WM-001] World Models](../vol05_agi/AGI-WM-001_world_models.md): Internal models for planning.

## 11. References & Foundational Literature
1. Levine, S., Finn, C., Darrell, T., & Abbeel, P. (2016). *End-to-End Training of Deep Visuomotor Policies*. Journal of Machine Learning Research, 17(39), 1–40. https://arxiv.org/abs/1504.00702. [End-to-end training of perception and control from images to torques.]
2. Tobin, J., Fong, R., Ray, A., Schneider, J., Zaremba, W., & Abbeel, P. (2017). *Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World*. IROS 2017. https://arxiv.org/abs/1703.06907. [Randomised simulation as a route across the reality gap.]
3. OpenAI, Akkaya, I., Andrychowicz, M., Chociej, M., Litwin, M., McGrew, B., Petron, A., Paino, A., Plappert, M., Powell, G., Ribas, R., Schneider, J., Tezak, N., Tworek, J., Welinder, P., Weng, L., Yuan, Q., Zaremba, W., & Zhang, L. (2019). *Solving Rubik's Cube with a Robot Hand*. arXiv:1910.07113. [Automatic domain randomization and simulation-only training for a robot hand.]
4. Lee, J., Hwangbo, J., Wellhausen, L., Koltun, V., & Hutter, M. (2020). *Learning Quadrupedal Locomotion over Challenging Terrain*. Science Robotics, 5(47), eabc5986. https://doi.org/10.1126/scirobotics.abc5986. [Simulation-trained locomotion policy on a legged robot.]
5. Zhao, W., Queralta, J. P., & Westerlund, T. (2020). *Sim-to-Real Transfer in Deep Reinforcement Learning for Robotics: a Survey*. IEEE Symposium Series on Computational Intelligence (SSCI 2020), 737–744. https://doi.org/10.1109/SSCI47803.2020.9308468. [Survey of reality-gap methods.]
6. Kaufmann, E., Bauersfeld, L., Loquercio, A., Müller, M., Koltun, V., & Scaramuzza, D. (2023). *Champion-level drone racing using deep reinforcement learning*. Nature, 620(7976), 982–987. https://doi.org/10.1038/s41586-023-06419-4. [Swift drone-racing system and head-to-head races against human champions.]
7. Ahn, M., Brohan, A., Brown, N., et al. (2022). *Do As I Can, Not As I Say: Grounding Language in Robotic Affordances*. arXiv:2204.01691 (45 authors). [Grounding language-model plans with learned skill value functions.]
8. Driess, D., Xia, F., Sajjadi, M. S. M., et al. (2023). *PaLM-E: An Embodied Multimodal Language Model*. ICML 2023. https://arxiv.org/abs/2303.03378. [Embodied multimodal language model for robotics.]
9. Brohan, A., Brown, N., Carbajal, J., et al. (2023a). *RT-1: Robotics Transformer for Real-World Control at Scale*. Robotics: Science and Systems 2023. https://arxiv.org/abs/2212.06817. [Transformer policy trained on large-scale real-robot data; scaling with data and model size.]
10. Brohan, A., Brown, N., Carbajal, J., et al. (2023b). *RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control*. arXiv:2307.15818. [VLA co-fine-tuning on web and robot data; 6,000 evaluation trials.]
11. Open X-Embodiment Collaboration, O'Neill, A., Rehman, A., et al. (2024). *Open X-Embodiment: Robotic Learning Datasets and RT-X Models*. ICRA 2024. https://arxiv.org/abs/2310.08864. [Pooled multi-robot dataset (22 robots, 21 institutions) and positive-transfer result.]
12. Kim, M. J., Pertsch, K., Karamcheti, S., et al. (2024). *OpenVLA: An Open-Source Vision-Language-Action Model*. arXiv:2406.09246. [Open 7B VLA and its reported comparison to RT-2-X.]
13. Black, K., Brown, N., Driess, D., et al. (2024). *$\pi_0$: A Vision-Language-Action Flow Model for General Robot Control*. arXiv:2410.24164. [Flow-matching VLA trained on multiple robot types.]
14. Chi, C., Xu, Z., Feng, S., Cousineau, E., Du, Y., Burchfiel, B., Tedrake, R., & Song, S. (2023). *Diffusion Policy: Visuomotor Policy Learning via Action Diffusion*. RSS 2023 (arXiv:2303.04137). [Diffusion-based visuomotor policies and the reported 46.9% average improvement.]
15. Zhao, T. Z., Kumar, V., Levine, S., & Finn, C. (2023). *Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware*. RSS 2023 (arXiv:2304.13705). [Action Chunking with Transformers on low-cost hardware.]

## 12. Evidential Classification Matrix
| Section / Claim | Evidence Category | Supporting Source | Confidence / Status |
| :--- | :--- | :--- | :--- |
| Robot control as partially observable sequential decision-making | Established | Standard formulation; Levine et al. (2016) | High |
| Behaviour cloning suffers compounding errors; action chunking mitigates | Well Supported | Zhao et al. (2023) | Moderate |
| Simulation-trained policies can transfer to real robots in specific tasks | Well Supported | OpenAI et al. (2019); Lee et al. (2020); Kaufmann et al. (2023) | Moderate to High |
| Swift matched human champions in racing | Well Supported (single group, peer reviewed, specific course/conditions) | Kaufmann et al. (2023) | Moderate |
| VLA models improve instruction/object generalisation | Preliminary (developer-reported) | Brohan et al. (2023b); Kim et al. (2024) | Moderate |
| Cross-embodiment positive transfer | Preliminary to Well Supported | Open X-Embodiment Collaboration (2024) | Moderate |
| Diffusion policy outperforms prior methods | Well Supported on tested benchmarks | Chi et al. (2023) | Moderate |
| General-purpose household robots | Forecast | `FCST-ROBOT-001` | Low to Medium |

## 13. Revision History
- **v0.1.0** (2026-10-15): Initial draft. Citations checked against arXiv/Crossref/publisher pages where marked; not independently fact-checked.