Ken Goldberg, UC Berkeley professor and AUTOLab lead and Ambi Robotics co-founder, discusses agentic robotics and production-ready automation with The Humanoid Hub at CoRL in Austin.
Goldberg says he had been skeptical because robotics lacks the vast amounts of training data available in other fields. His view changed when vision-language models such as Astra began generating robot code, including control for cameras and grasping. He says the code can be inspected and run directly on production robots, without relying on a black-box policy.
For industrial use, Goldberg prioritizes speed, reliability, and interpretability; he says systems that are slow or work only 80% of the time are unlikely to be adopted on production lines. His bottom-up approach is to solve specific tasks, deploy them, then learn from production data. He says agentic tools could cut engineering setup that often takes six weeks down to a day or two, opening opportunities for smaller companies.
Goldberg names three projects: Code as Policy (CaP), CaP-X, and Graph as Policy (GaP), whose code is open for others to try. He also describes RoboSimStudio, which creates a physics-based Newton simulation from real-world information, including material properties, friction, and deformation. The resulting ROS 2 code can run on a robot.
Goldberg sees agents assembling established, trusted software skills such as edge detection and segmentation. Engineers still need to understand these components, while agents can help configure them for a particular environment. He calls this a return to practical engineering methods that are fast and inspectable.
Goldberg distinguishes home robots, where broad abilities may be useful, from factory robots that can specialize in particular jobs. He argues that simple arms and grippers can handle many industrial tasks with less weight, cost, and risk of failure than complex hands. He expects solutions for repetitive work such as fruit sorting and food assembly to build toward greater generality, while saying humanoids may make more sense in homes than factories.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Bloomberg Originals host and narrator, with 1X founder Bernt Børnich, UC Berkeley professor Ken Goldberg, GXO’s Patrick Kelleher, and other robotics, industry, and labor experts, examines humanoid robots’ potential and the hurdles they face in factories, warehouses, homes, and hospitals.
Ken Goldberg, a professor at UC Berkeley, discusses how robotics can combine learned models with engineering at an IEEE Robotics and Automation Society event.
Ken Goldberg, a robotics researcher at UC Berkeley and co-founder of Ambi Robotics, explains at TEDxMarin why household robots struggle with manipulation and describes progress in warehouse sorting and home tasks.
Ken Goldberg, a roboticist at UC Berkeley, presents research on robot grasping for the University of Toronto Robotics Institute, hosted by his former PhD advisor Animesh.