UC Berkeley Professor Ken Goldberg shapes the future of robotics | AI In Motion

Robotics advancement relies on balancing traditional engineering principles with modern data-driven AI. While large models show promise, fundamental tools like PID controllers and inverse kinematics remain essential for bootstrapping systems and ensuring modularity. The field faces a significant "data gap," particularly in manipulation, where complex contact dynamics and material deformations defy simple simulation. Strategic data curation—focusing on high-quality, bottleneck-specific samples rather than brute-force collection—is crucial for progress. Furthermore, computer vision is increasingly serving as a proxy for tactile sensing, with technologies like 3D Gaussian splats enabling high-fidelity modeling of real-world environments. Ken Goldberg, a professor at UC Berkeley, emphasizes that these infrastructure-level innovations, rather than just raw scale, provide the path to robust, real-world autonomous systems, ranging from surgical robotics to self-driving vehicles.