Boston Dynamics' Pursuit of Natural and Elegant Robot Movement: A Conversation with CEO Robert Playter

This episode features Robert Playter, CEO of Boston Dynamics, discussing the company's three-decade journey in creating advanced legged and humanoid robots like Atlas and Spot. Playter, who started at MIT's legendary Leg Lab, delves into the deep fascination with movement that drives their work, emphasizing the challenge and beauty of achieving natural, dynamic locomotion in machines. He highlights the long and arduous process of developing control algorithms, noting that truly natural walking for humanoids like Atlas took 10-15 years to perfect, often emerging as a byproduct of new development processes.

The conversation explores the core principles guiding Boston Dynamics, including simplifying complex problems to their essence (e.g., starting with a pogo stick to understand dynamic balance), working with the physics of the machine rather than overpowering them, and fostering a culture of fearless iteration where breaking and repairing robots is seen as essential for progress. Playter distinguishes between the 'science' of developing principles and the 'art' or intuition involved in translating human-like movement into robots, especially for complex forms like humanoids. He explains the challenges posed by singularities in joint movement and the under-actuated nature of legged robots, where external forces are only mediated through foot placement.

Playter details the specific difficulties of humanoid robotics, such as managing the inertia of a large upper body and handling heavy objects, which significantly shifts the robot's center of mass. He proudly notes Boston Dynamics' differentiation in enabling Atlas to manipulate substantial weights, a capability often lacking in other humanoid projects. A significant breakthrough is the rapid acceleration in behavior development, with new motions now achievable in days rather than months, thanks to advanced model predictive control and optimization techniques.

Looking ahead, Playter anticipates a future where robots like Atlas will surpass human physical capabilities in certain movements, such as jumping height, due to increasingly powerful actuators. The episode underscores the profound satisfaction and happiness experienced by roboticists in bringing machines to life and the enduring value of pursuing curiosity and tackling multi-decade problems, a lesson learned from his mentor, Mark Raibert. The discussion ultimately paints a picture of robotics as a field where scientific rigor, engineering ingenuity, and an artistic appreciation for movement converge to push the boundaries of what machines can achieve.

I probably didn't really see the nice natural walking that I expected out of our humanoids until maybe last year.

The fascination the Deep Fascination is really about movement.

How do you build something so that the physics of the machine just kind of inherently wants to do what it wants to do?

It's okay to pursue your interest your curiosity do something because you love it.

Try to simplify a thing to the Core Essence.

It's okay it's good to break stuff.

The insight of dynamic stability and a lighted machine is to go with it you know let the Tipping happen you know let yourself fall but then catch her catch yourself with that next foot.

When your leg is fully extended it can't go further the other direction right there's only you can only move in One Direction and that makes all of the calculations around how to produce twerks at that joint or positions makes it more complicated.

The humanoid form is attractive in many ways but it's also a challenge in many ways.

We were able to create in a matter of days. It used to be it took six months to do anything new.

The robots the actuators are going to get stronger where there really is the case that at least some of these joints some of these motions will be stronger.

We're running these model predictive control techniques where we're able to the robot essentially can think in advance for the next second or two about how its motion is going to transpire.