Registration- you MUST RSVP and register to enter this event, there will be a manual check-in outside of Kiva (32-G449): https://www.tesla.com/event/tesla-x-mit-csail
When/Where: Wednesday, October 29th in CSAIL Kiva Seminar Room (32-G449) from 2:00-4:30pm
Registration Required: https://www.tesla.com/event/tesla-x-mit-csail
Agenda:
2:00 – 3:30 PM | Technical Talks from Tesla Engineering Team:
Talk 1 (2-2:30pm): Solving Self-Driving with a Modern AI Approach
- Abstract: This talk will present the approach Tesla is using to solve self-driving at scale. We will discuss the end-to-end technology used to train the driving policy model. We will cover how we curate the largest embodied AI dataset in the world from the millions of cars in our fleet. And we will highlight how your work can contribute and make real-world impact.
- Speaker: Lawson Fulton is a Senior Machine Learning Engineer at Tesla Autopilot, where he trains production neural networks for over 5 million Tesla customers. Lawson works on a team of engineers and scientists building Autopilot’s world reconstruction models. Previously, he completed his master's at the University of Toronto in machine learning and computer graphics.
Talk 2 (2:30-3:00pm) : Building close-loop simulation with real time
generative modeling
- Abstract: As end-to-end policies get closer to human-level performance, it becomes an increasingly bigger challenge to identify edge / failure cases in the wild and reproduce them reliably. Simulation that can consistently evaluate the policies's performance against certain edge cases therefore becomes critical for model development. In addition, such system should be sufficiently versatile to apply to any scenario, support control-in-the-loop for policy fidelity, and have low compute requirement for scalability. Given these requirements, we developed a real-time, close-loop simulation system based on generative modeling, and we will demonstrate how such system allows us to reproduce real-world interventions in a generated world.
- Speaker: Zhichun (Eric) Huang received his master degree in machine learning in 2021 from CMU, where he collaborated with Prof. Zico Kolter on applications of deep equilibrium models. He joined Tesla Autopilot in 2022, leading effort in scaling generative models for perception, end-to-end control and simulation.
Talk 3 (3-3:30pm) : Building a general-purpose Humanoid Robot
- Abstract: Tesla is on a mission to redefine the productivity of our civilization through the development of intelligent, general-purpose humanoid robots. In this talk, we will share our vision and progress toward building scalable robotic systems that can perform a wide range of tasks with human-level capability. We will review the major milestones achieved in hardware design and embodied AI, and present our strategy for developing productionizable, general-purpose robotic intelligence. This includes our approach to scalable learning, simulation, and real-world deployment. Attendees will gain insight into how Tesla is leveraging its expertise in AI and manufacturing to accelerate the future of robotics and transform the way we live and work.
Speaker Andy Tsai is a member of the Tesla AI team, focused on advancing general-purpose robotic intelligence. He built his first robot during his study at the Robotics Institute at Carnegie Mellon University and brings over a decade of experience in robotic AI to his work
3:30 – 4:30 PM | Q&A, Networking, Optimus Demo
