# AI with a body: robots that learn · AI in Bed

Advanced level · 8 min · https://www.aiinbed.com/learn/advanced/ai-and-robotics

*Why folding laundry is harder than passing exams, and how robots are starting to learn from simulation, demonstrations and language models.*

Chatbots live in text. Robots live in the messy physical world, where things slip, lights change and a mistake can break a glass. AI that senses and acts in the world is called **embodied AI**.

### Sense, plan, act

**Perception**: turn camera images, depth readings and touch sensors into an understanding of what is where.

**Planning**: decide what to do, from "pick up the cup" down to a path for the arm.

**Control**: send precise, rapid commands to the motors, correcting many times a second as things move.

### Moravec's paradox

In the 1980s, Hans Moravec and others noticed something odd: computers find chess and algebra easy but struggle with things a toddler does, like walking or grasping a toy.

**Moravec's paradox** has a likely explanation. Evolution spent hundreds of millions of years polishing our senses and movement, so they feel effortless. Abstract reasoning is recent, so it only feels hard to us.

**Quick check:** According to Moravec's paradox, which is hardest for a robot?

A. Multiplying two large numbers
B. Playing a good game of chess
C. Picking a soft, crumpled towel out of a messy basket
D. Sorting a list of names

Answer: C. Everyday handling of floppy, unpredictable objects needs rich perception and fine control, which is exactly what machines find hard.

### Where robots get their data

Chatbots learn from the internet. There is no internet of robot movements, so data is precious.

**Simulation**: robots practise millions of times in a virtual world. The gap between simulation and reality is the **sim-to-real** problem. One fix is domain randomisation: vary colours, friction and lighting in the simulator so the real world looks like just another variation.

**Learning from demonstrations**: a person guides the robot or controls it remotely, and the robot learns to imitate those movements. This is called imitation learning.

> Learning to ride a bike from a video game helps a little, but the real wobble always feels different. That gap is sim-to-real.

### Vision-language-action models

A **vision-language-action model**, or VLA, starts from a model trained on images and text from the web, then is trained further on robot data. It takes in a camera view and an instruction, and outputs robot actions, often written as special tokens.

Google DeepMind's RT-2, in 2023, showed this lets robots use web knowledge, for example following instructions about objects they never saw during robot training.

- Data is scarce and slow to collect, one real movement at a time.
- Mistakes are physical: they can break things or hurt people, so safety matters more.
- Control must happen in real time; a robot cannot pause to think mid-grasp.
- Every home and warehouse is different, so robots must cope with endless variety.

**Quick check:** What is the sim-to-real problem?

A. Robots refusing to leave the simulator
B. Skills learned in simulation working less well on real hardware because the virtual world is not quite real
C. Simulators being too expensive to run
D. Real robots being faster than simulated ones

Answer: B. Small differences in physics, friction, lighting and sensors mean behaviour that works in simulation can fail in reality.

**Key takeaways**

- Embodied AI must perceive, plan and control in the real world.
- Moravec's paradox: everyday movement is harder for machines than abstract thought.
- Simulation, demonstrations and VLA models are how robots are catching up.

**Go deeper (free sources)**

- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://arxiv.org/abs/2307.15818) by Brohan et al., Google DeepMind
- [Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World](https://arxiv.org/abs/1703.06907) by Tobin et al.
- [Open X-Embodiment: Robotic Learning Datasets and RT-X Models](https://arxiv.org/abs/2310.08864) by Open X-Embodiment Collaboration
- [LeRobot](https://github.com/huggingface/lerobot) by Hugging Face