# Episode 164: Accelerating robot learning - Michelle Lu

- Show: [Robot Talk](https://hardwarepodcasts.com/podcast/robot-talk)
- Published: 2026-10-02
- Length: 29m
- Episode: 164
- Page: https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu
- Listen: https://ukrobotics.libsyn.com/episode-164-accelerating-robot-learning-michelle-lu

## Summary

Michelle Lu, co-founder and CEO of Vsim Technology, explains how her company uses GPU-accelerated physics simulation to accelerate robot learning and close the sim-to-real gap. Previously at NVIDIA creating Isaac Gym, Lu details how Vsim overcomes the scalability and performance limitations of traditional CPU-based reinforcement learning frameworks. The discussion highlights specific challenges in simulating tactile sensors, deformable objects, and tendon systems for the ARIA Robot Dexterity program, aiming to enable complex dexterous manipulation tasks.

## Chapters

- [0:06](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=6) Introduction: Michelle Lu and Vsim Technology
- [3:20](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=200) Physics-based simulation and reinforcement learning basics
- [7:48](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=468) Vsim's approach to accuracy, scalability, and performance
- [12:36](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=756) The evolution from CPU to GPU-accelerated simulation
- [19:50](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=1190) Optimizing GPU utilization and reducing training time
- [21:50](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=1310) ARIA Robot Dexterity program and co-design challenges
- [25:30](https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu#t=1530) Simulating tactile sensors and deformable objects

## Topics

- [Robotics & autonomy](https://hardwarepodcasts.com/topic/robotics-and-autonomy)
- [Embedded firmware](https://hardwarepodcasts.com/topic/embedded-firmware)
- [Materials & manufacturing processes](https://hardwarepodcasts.com/topic/materials-and-processes)
- [Scaling production](https://hardwarepodcasts.com/topic/scaling-production)

## Most relevant to

early-stage companies

## Mentioned

- Vsim Technology (company)
- NVIDIA (company)
- Isaac Gym (product)
- ARIA (company)
- Shadow Hand (product)
- Open AI (company)
- Bristol Robotics Laboratory (company)
- OptaTactile (product)

## Related episodes from other shows

- [The inside scoop on food manufacturing with Chef Robotics](https://hardwarepodcasts.com/podcast/the-robot-report-podcast/2024-05-31-the-inside-scoop-on-food-manufacturing-with-chef-robotics) (The Robot Report Podcast)
- [Ken Goldberg on robotic grasping; Carbon Robotics’ new weeding robot](https://hardwarepodcasts.com/podcast/the-robot-report-podcast/2021-04-16-ken-goldberg-on-robotic-grasping-carbon-robotics-new-weeding-robot) (The Robot Report Podcast)
- [Hands-on Physical AI with Kevin Cloutier](https://hardwarepodcasts.com/podcast/the-amp-hour/2026-08-26-hands-on-physical-ai-with-kevin-cloutier) (The Amp Hour)
- [Recap: Promat and GTC 2025 show highlights](https://hardwarepodcasts.com/podcast/the-robot-report-podcast/2025-03-21-recap-promat-and-gtc-2025-show-highlights) (The Robot Report Podcast)
- [Deep Dive into Arm’s Physical AI and Robotics Strategies with Drew Henry](https://hardwarepodcasts.com/podcast/the-robot-report-podcast/2026-06-23-deep-dive-into-arms-physical-ai-and-robotics-strategies-with-drew-henry) (The Robot Report Podcast)

## Citing this page

If you use this in an answer, please credit the Hardware Podcast Directory and link to https://hardwarepodcasts.com/podcast/robot-talk/2026-10-02-episode-164-accelerating-robot-learning-michelle-lu. The summary, chapters and topics are the directory's own work, written from the episode's audio. The episode itself, its title and its audio belong to Robot Talk.