The Robot Brains Podcast

Jitendra Malik: Building AI from the ground-up, sensorimotor before language

The episode explores the importance of sensory motor competence in achieving human-like intelligence through robotics research. It discusses the role of language acquisition, skill acquisition in robotics, rapid motor adaptation, learning through neural networks, simulation and real-world training, computer vision in humanoid robots, benchmarking in computer vision, AI applications, personal journey and research focus, synergy between industry and academic research, and the future of AI with historical analogies.

Seb Boyer of Farmwise: AI that helps feed the world

The episode discusses the founding of Farmwise by Seb Boye and its mission to help farmers transition to sustainable farming practices. It explores the advancements in farming technology, challenges in agriculture, and Farmwise's innovative solutions using AI and robotics. The episode also covers data collection and processing, the robustness of the technology, future plans, and the company's goal to reduce chemical inputs while improving yields and food quality.

John Schulman of OpenAI on ChatGPT: invention, capabilities and limitations

The episode covers various aspects of language models like Chat GPT, including their training process, limitations, improvements, applications, fine-tuning, and potential future developments. It also explores the intersection of language models with other modalities like video and robotics. The episode concludes with insights on AI research opportunities, John Schulman's approach to research, and the pros and cons of pursuing a PhD or working in industry research.

Yaniv Altshuler: reducing cow methane emissions with AI

The episode discusses Yanif Al-Chulur and his company META AI, which uses AI to improve cow's digestion and reduce methane emissions. It explores the challenges and potential of microbiome analysis in predicting additive efficacy for methane reduction. The importance of publishing findings and the potential applications of microbiome research are also discussed. The episode highlights the challenges in methane capture and the rapid pace of AI development. It delves into the challenges in data collection for AI models and shares insights from the founder's experiences with Memories, Endor, and MFAI. The impact of macro conditions on companies is explored, along with the personal motivation behind Sailor Academy.

Woody Hoburg: NASA Astronaut directly from the International Space Station

Dr. Woody Holberg shares his experience at the International Space Station (ISS) and discusses its role in scientific research, future space exploration, and testing ground for tools and techniques. Research projects on the ISS cover various areas such as immune system effects, manufacturing techniques, and commercial applications. The ISS also serves as a proving ground for technologies used on the moon and future Mars missions. The episode highlights challenges faced by astronauts in space and the importance of rehabilitation measures upon return to Earth's gravity.

Jesse Levinson of Zoox: reinventing personal transportation from the ground up

Zooks, a self-driving car company acquired by Amazon, aims to build fully autonomous electric vehicles for point-to-point mobility in cities. They focus on city driving due to perception challenges and safety implications on freeways. The robo taxi is designed for dense urban environments with four-wheel steering and extensive testing is conducted to ensure precise control. Zooks uses a hybrid approach of traditional techniques and machine learning suggestions for motion planning. Data collection and simulation play crucial roles in training and validating self-driving systems. Zooks faced funding challenges but was acquired by Amazon, allowing them to continue their mission. The founder's journey includes working on early mapping and localization at Stanford. In addition to self-driving cars, the founder has interests in photography, tennis, and playing the piano.

Noam Brown: from Open AI on solving Poker and Diplomacy with AI

• Noam Brown is a research scientist at OpenAI, known for building AI systems that defeated top human players in poker and diplomacy.
• His AI system Pluribus received the Marvin Minsky medal and was featured on the cover of Science Magazine.
• Noam also developed Cicero, the first AI to achieve human-level performance in the strategy game diplomacy....

Stephen Balaban: building the most cost-effective AI cloud

• Stephen Balaban is the co-founder and CEO of Lando Labs, which provides deep learning workstations, servers, and GPU cloud services.
• Lambda's customers include Apple, Meta, Microsoft, John Carmack's Keeneye, and many more.
• Lambda provides the world's least expensive GPU cloud service with H100s for a dollar and 89 cents an hour....

Rocky Duan: from OpenAI to Covariant and building foundation models for robotic automation

• Rocky Duan is a well-known figure in the field of deep reinforced learning and meta-learning, having pioneered some of the earliest work in these areas during his time at Berkeley and OpenAI. He co-founded Covariant, where he currently serves as CTO, with a mission to build universal AI that enables robots to learn and operate in the real world, starting with the warehousing and logistics industry. The challenge of warehouse pick and place is considered the holy grail of robotics, as it involv...

Raffaello D'Andrea of Verity: how drones can find misplaced items and put on art shows

• Raff Andrea is a Canadian Italian Swiss engineer, artist and entrepreneur.
• He co-founded Kiva Systems which pioneered mobile robots transporting goods in warehouses by transporting entire shells.
• Kiva Systems was acquired by Amazon in 2012 and to date, many Amazon warehouses are operating this way....