RoboBoom
The Humanoid Hub· October 10, 2025· 36:09

The Next Chapter for Atlas | The Promise of Humanoids with Mario Bollini of Boston Dynamics

Mario Bollini, who leads Atlas product management at Boston Dynamics, discusses the robot’s development and commercial prospects with the Humanoid Hub.

In this talk

  1. 0:13

    From robotics projects to product management

    Bollini traces his interest in building things from Science Olympiad and Rube Goldberg machines to mechanical engineering at MIT, where he added computer science and electrical engineering. In graduate school, he worked on autonomous cooking with the PR2, then helped start companies making outdoor wheelchairs and plantable pencils. He joined Boston Dynamics as the company began commercializing Spot.

  2. 2:18

    What Spot taught about selling robots

    Bollini says Spot’s early adopter program helped the team find a niche in dynamic sensing, facility maintenance, and monitoring. A customer’s excitement about buying a first robot is not proof that it delivers value. Successful adoption requires support from multiple stakeholders, including operators and maintenance teams, and fitting the robot into a company’s broader digital transformation.

  3. 4:08

    Why Atlas moved from hydraulics to electric

    Bollini says battery energy density made electric Atlas feasible, and Boston Dynamics developed its own motors and actuators to meet performance needs. Hydraulic Atlas was powerful but ran at thousands of psi, could leave fluid on floors, and was difficult to maintain. Electric actuation was a necessary step toward a robot that could be sold to customers.

  4. 5:32

    New control methods and product trade-offs

    The earlier Atlas relied on classical control approaches such as model predictive control. The newer program has adopted reinforcement learning for locomotion and behavior cloning for manipulation, with partners including RAI, NVIDIA, and TRI. Bollini says learned whole-body policies have shortened development time, while actuator choices are guided by cost, customer return on investment, and requirements derived from target tasks.

  5. 9:36

    Designing for flexible work cells

    Atlas’s rotating leg and broad joint range were deliberate engineering choices that can let the robot turn in place and potentially reduce cycle times. Bollini says the team is evaluating tasks such as loading parts, removing tires from trailers, and moving debris at Fukushima. He sees the humanoid form as a flexible way to test commercial opportunities, while allowing that some tasks may eventually suit other robot designs better.

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

More talks