Interview, Fireside Chat
Robert Playter: Boston Dynamics CEO on Humanoid and Legged Robotics | Lex Fridman Podcast #374
- Development of new walking control algorithms for the next generation of Atlas is expected to yield natural-looking motion as a by-product over a 10 to 15-year timeframe spanning from the 2008 Pet Man Prototype to 2022, with future approaches becoming broadly available and utilizing model predictive control to accelerate from six-month to days-long behavior creation timelines.
- Next-generation Atlas actuators are projected to be stronger, enabling physical feats like box jumps beyond human capability, while dexterity focuses on two-armed manipulation for unwieldy objects in manufacturing or construction, complemented by improved real-time calculations from more capable computers.
- Implementation of behaviors on soft surfaces like sand is anticipated to be achievable despite requiring specific implementation time, and the "beauty" of movement derived from natural physics usage is identified as a key differentiator for future machines.
- Physical indication layers using body movement to communicate intent will gain importance for human-robot interaction as robots enter real-world environments, supported by commoditized vision algorithms available as third-party downloadable APIs.
- The path to consumer-level social robots involves an initial phase of industrial applications and mass manufacturing to reduce costs and improve reliability, potentially leading to products capable of broad tasks such as fetching beverages or cleaning dishes.
- Generative AI capabilities, including conversational chat, are expected to be integrated into the Spot platform faster than anticipated, possibly leveraging open-source alternatives.
- The Boston Dynamics AI Institute plans to pursue "over the horizon" research problems with horizons extending beyond five years, free from commercial constraints, while the industry anticipates a proliferation of robotic companies driven by sector validation.
- Long-term expectations include the fear of job loss being mitigated by upskilling workers as robot operators and the emergence of ubiquitous robots performing hard labor complementarily, with humans retaining creative work, alongside potential future robots serving as emotional companions to address loneliness.