Robot Talk · 22 May 2026 · 28m · Episode 157
Josie Hughes, Assistant Professor at EPFL and founder of the CREATE Lab, discusses using generative AI to design robotic manipulators and soft robots. She explains how her team combines large language models with simulation and reinforcement learning to generate task-specific gripper morphologies, controllers, and part selections. The conversation covers the challenges of spatial reasoning in AI, the trade-offs between specialized and general-purpose designs, and the potential of morphing robots that can adapt their physical form to different environments.
Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Design
The team uses large language models to generate initial gripper morphology, controller, and part selection designs from text or video inputs, which are then refined through simulation and reinforcement learning loops.
“we're taking some initial design ideas from our AI reasoning element, and we have that in a wider pipeline, and we're combining this with simulation and reinforcement learning to do what we're calling local optimization.”
Josie Hughes · Hear it at 11:31
early-stage companies
Other shows on the same ground, closest first.
Dinesh Narayanan, Head of Commercialization at General Robotics, explains the company's transition from Microsoft Research to build Grid, a platform integrating simulation, AI models, and deployment pipelines to accelerate robotics development. He details how Grid’s modular 'plug-and-play' skills allow enterprises to rapidly prototype and deploy solutions across diverse form factors, including humanoids, by combining various AI techniques rather than relying on a single monolithic model. The conversation highlights the platform's ability to handle complex workflows, such as emergency kit packing, by orchestrating specialized models for vision, manipulation, and navigation simultaneously. Narayanan positions General Robotics as the 'AWS for robotics,' focusing on deployment speed and adaptability for large enterprises and government organizations.
Joe Bogusky, a Senior Mechanical Engineer at Intuitive Surgical, discusses his 25-year career developing minimally invasive medical devices, ranging from robotic-controlled catheters to the DaVinci surgical robot. He details the unique engineering challenges of catheter design, such as managing conflicting stiffness requirements and testing materials like PTFE in simulated body temperature environments. Bogusky also shares insights on the future of surgical robotics, emphasizing the role of AI for low-risk tasks while maintaining physician control, and advises engineers to build custom calculation tools and study biomimicry for inspiration.
Rajat Bhageria, founder and CEO of Chef Robotics, explains how his company automates food assembly in high-volume manufacturing plants using AI-enabled robots with interchangeable titanium grippers. He details the challenges of manipulating deformable, sticky food items where simulation fails, requiring real-world data collection to train models for consistent portioning and handling. The episode also features Joyce Sidopoulos, co-founder and COO of MassRobotics, discussing their no-equity accelerator program for robotics startups and initiatives to support women in STEM.
Milan Kocic, Founder and Head of Sixth Sense at Hexagon Manufacturing Intelligence, discusses the role of AI in advanced manufacturing and the challenges of the US workforce. He highlights the Sixth Sense accelerator program, which supports startups like GelSight, and explains how generative AI tools from companies like Rafinex and Caster are augmenting design engineers rather than replacing them. Kocic argues that the future of manufacturing lies in collaborative robotics and AI-driven efficiency to address labor shortages and reshape the industry's public perception.
Matthias Wagner, Founder and CEO of Flux, explains how his company uses agentic AI to democratize PCB design, allowing non-engineers to create hardware via natural language prompts. He discusses the historical bottlenecks of hardware design, the necessity of iterative back-and-forth for complex devices like smartphones, and the economic advantages of designing custom boards for products like vending machines. Wagner also outlines future integration challenges, including firmware generation and enclosure design, and predicts that lowering the cost of iteration will unlock significant innovation in robotics and IoT.
The summary, chapters and topics are generated from the episode’s audio by this directory, not written by Robot Talk. The episode itself belongs to its publisher.