Robot Wiki

Data, Hardware & Evaluation

Data, hardware and evaluation are what robot learning runs on: training data, the robots that collect it and the tests that score results. In 2025, 5 million industrial robots were in operation International Federation of Robotics 2026.

Domain overview: 7 articles. Data first, then hardware, evaluation and deployment.

Related: Market Map, Robot Learning Curriculum for ML Engineers, The Reliability Gap

  1. The Data Bottleneck

    Robot data comes in hours; language data in trillions of tokens.

  2. Major Datasets

    The major open robot datasets and what each collected.

  3. Hardware Taxonomy

    A buyer's guide from SO-101 to Jetson Thor.

  4. Teleoperation Rigs

    ALOHA, GELLO, UMI and VR rigs compared on cost and quality.

  5. The Evaluation Crisis

    What a success rate hides, from trial count to scorer.

  6. Industrial Deployment

    The jam-rate arithmetic that decides whether a 99 percent cell ships.

  7. The Robot Learning Stack

    From data capture to robot integration as one reproducible system.