In this episode of the Google DeepMind podcast, host Hannah Fry interviews Carolina Parada, the lead of robotics research at Google DeepMind, about the evolution and future of AI in robotics. Parada discusses the advancements made, including the introduction of LLMs, VLMs, and Gemini Robotics, which brings multimodal understanding to robots. The conversation covers the challenges of achieving general-purpose robots, the importance of embodied reasoning, and the integration of Gemini's capabilities for complex tasks like packing lunches and even performing a slam dunk. Parada also addresses safety concerns, the Asimov dataset, and the potential for robots to collaborate with humans in various fields, highlighting the breakthroughs needed for more advanced and socially intelligent robots.
