---
title: "AI Designed 12,774 Robots, but Reality Proved Tougher Than Simulation: MIT Built 135 Machines and Achieved 66% Success"
description: "MIT AI agents designed 12,774 robots and ran 16,427 virtual tests, but only 66% of hand-built machines coped with real-world obstacles."
date: 2026-11-20T23:03:50.040Z
lang: en
url: https://xab.info/en/posts/ai-designed-robots-mit-66-percent-success-reality
tags: [ai-robots, mit, robotics, simulation, 3d-printing, machine-learning]
author: "Alexander Buran, Roman V."
publisher: "XAB.info"
---

# AI Designed 12,774 Robots, but Reality Proved Tougher Than Simulation: MIT Built 135 Machines and Achieved 66% Success

![Visualization of numerous unique robots designed by AI agents to overcome an obstacle course.](https://xab.info/media/2026/11/20/ii-sproektiroval-12774-robotov-mit-66-uspeha-v-realnosti/ii-sproektiroval-12774-robotov-mit-66-uspeha-v-realnosti-1.webp)

## Key Points

- Four AI agents in the RoboBuilder system designed 12,774 robots and conducted 16,427 virtual tests
- MIT scientists manually built 135 prototypes in ~240 hours; 66% coped with real obstacles
- After a video feedback loop, the success rate on the first two levels rose from 73% to 93%
- AI invented mechanisms not found by automated search algorithms: lugged wheels, linkages, cable and walking systems
- On complex obstacles and the 'one robot for the whole track' task, AI performed significantly worse

Researchers at the Massachusetts Institute of Technology (MIT) have presented the results of a large-scale experiment in which four AI agents within the RoboBuilder system independently designed 12,774 robot structures and conducted 16,427 virtual tests. From the designs that successfully passed simulation, scientists manually built 135 physical prototypes, spending about 240 hours on manufacturing. However, only 66% of them coped with real obstacles across all difficulty levels, clearly demonstrating the gap between digital modeling and the physical world.

### How the RoboBuilder System Works

The RoboBuilder system allows AI agents to design not only software but also the physical construction of robots. Agents selected parts from a ready-made component set, picked motors, created additional elements for 3D printing, and wrote motor control programs. Each agent's task was to overcome a track of 10 obstacles, each having three difficulty levels ranging from basic to advanced. Four AI models participated in the experiment, each running its own design and optimization pipeline.

### The Gap Between Simulation and Reality

The main problem emerged during the transition from virtual space to the physical world. Tires slipped on surfaces, component joints came apart under load, gears disengaged, and the center of mass shifted when moving across inclined sections. Even after training the simulator on real measurements and video recordings, only 66% of structures that successfully passed virtual tests coped with physical obstacles at all difficulty levels. This means that more than a third of the robots that demonstrated flawless performance in simulation could not overcome even basic track elements in reality.

### Feedback Loop and Redesign Cycle

Researchers provided a feedback mechanism: if a robot failed an obstacle, the agent received a video recording of the failed attempt, analyzed it, and could change both the design itself and the simulator parameters. Following this refinement cycle, the share of successful real tests on the first two difficulty levels rose from 73% to 93%. This confirms that an iterative approach with visual feedback can significantly narrow the gap between virtual and physical robot behavior, although progress was less pronounced at the most complex levels.

### Unexpected Mechanisms and AI Capability Limits

Agents demonstrated the ability to invent mechanisms that automated search algorithms used in the experiment failed to produce. These include wheels with lugs for surface traction, special linkage structures, cable-movement mechanisms, and walking systems. Visual examples from the publication show that different models (GPT-6 Astra, Opus 5.5, GPT-6.1 Sol, Sonnet 5.5) generated fundamentally different approaches to the same obstacle: from rotating 'rubber legs' to cable trolleys and walking platforms with curved feet. At the same time, on complex obstacles and in the task of creating a single robot capable of completing the entire ten-element track, the AI performed significantly worse, pointing to current limitations in agents' ability to combine heterogeneous mechanisms into a single structure.

### Significance of the Experiment for Robotics

MIT results confirm that autonomous AI robot design is already capable of generating thousands of working concepts and finding non-trivial engineering solutions, but physical validation remains a critical stage that cannot be fully automated. The collected data on typical failures—slippage, joint separation, gear disengagement—will serve as valuable material for calibrating future simulators. The experiment also sets a clear benchmark: transitioning from virtual prototyping to mass production of AI-designed robots requires both improving physical models in simulators and developing automated assembly and diagnostic mechanisms.

## Fact-Check & Sources

- [ИИ спроектировал роботов для реальной полосы препятствий. Учёные MIT вручную собрали и протестировали 135 из них](https://www.ixbt.com/news/2026/11/20/ii-sproektiroval-robotov-dlia-realnoi-polosy-prepiatstvii-ucenye-mit-vrucnuiu-sobrali-i-protestirovali-135-iz-nix.html) - All key figures (12,774 structures, 16,427 tests, 135 built robots, 66% success, 240 hours, growth from 73% to 93%) are taken from the source text. Visual material confirms the presence of 4 AI models and the variety of designed mechanisms.

## FAQ

### What is the RoboBuilder system?

RoboBuilder is an MIT-developed system that allows AI agents to independently design both software and the physical construction of robots: selecting parts, picking motors, creating 3D printing elements, and writing control programs.

### Why did only 66% of robots cope with real obstacles?

When moving from simulation to reality, physical effects occurred that were not fully accounted for in the model: tire slippage, joint separation, gear disengagement, and load shifting on inclined surfaces.

### What happened after introducing video feedback?

The share of successful real-world tests on the first two difficulty levels rose from 73% to 93%. The agent analyzed the video of the failed attempt and could modify the design or simulator parameters.

### What mechanisms did AI invent on its own?

These include lugged wheels, special linkage structures, cable movement mechanisms, and walking systems. These solutions were not produced by automated search algorithms used in the experiment.