RoboBoom
IEEE Robotics and Automation Society· September 1, 2023· 1:00:39

IROS 35th Anniversary Forum Plenary 2: Marc Raibert -- A Culture of Robotics Research

Marc Raibert, robotics researcher and founder of the AI Institute, discusses Boston Dynamics’ robots and the research culture he wants to build at the IROS 35th Anniversary Forum.

In this talk

  1. 1:06

    Three roles for Boston Dynamics robots

    Raibert describes Spot as a general platform, Stretch as a machine designed for box handling, and Atlas as a testbed for exploring future capabilities. He says Spot is used for inspections and construction surveys, including radiation measurements at Fukushima. He reports nearly 900 Spots in use worldwide, with about 130 at academic institutions.

  2. 10:20

    Stretch and Atlas in practice

    Stretch targets the handling of cases in warehouses, a physically demanding job where employers struggle with turnover. Raibert traces its design back through Handle to Atlas and says customer pilots are underway. Atlas uses dynamic control and vision to adjust planned movements; he describes its control work as combining model predictive control, learned or practiced behaviors, perception, and engineering analysis.

  3. 17:58

    Hardware, cognition, and research culture

    Raibert argues that robotics advances when software, hardware, and physics are developed together. He says he started the AI Institute to pursue cognitive capabilities such as situational understanding, learning by watching, planning, and communication. Its research model aims to combine academic openness and long horizons with the teamwork and engineering depth of a corporate lab.

  4. 22:21

    Resilience and practical choices

    Raibert names ambition, diligence, persistence, and technical enjoyment as qualities he wants in robotics research. He illustrates diligence with robots continuing after disturbances, such as a shifted box or a pull on a rope. In the discussion, he says static or dynamic methods should fit the task, and that useful learning systems should be judged by whether they accomplish the intended job.

  5. 50:09

    Humanoids, collaboration, and public understanding

    Raibert says robot form should follow the application, while noting that humanoids attract greater public attention. He describes early work on making Spot’s actions clearer to people and says intimate physical assistance will require safer robots, better communication, and stronger cognitive abilities. He also supports public discussion of robotics risks and opportunities through an ethics and policy group at the AI Institute.

Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.

More talks