Embedded Files
November 9th, 2024
November 9th, 2024
08:50
Welcome and introduction
Welcome and introduction
Differentiable simulation
Differentiable simulation
09:00
Invited talk
Invited talk
Eric Heiden
Building High-Performance Differentiable Simulators in Warp
09:30
Invited talk
Invited talk
Felix Petersen
Differentiable Logic Gate Networks & Reinforcement Learning
10:00 Lightning talks
Poster session + coffee break 10:15 – 11:00
Improving simulation from data (1/2)
Improving simulation from data (1/2)
11:00
Invited talk
Invited talk
Yunzhu Li
Learning Structured World Models From and For Physical Interactions
11:30 Panel discussion I
What's the future of the interplay of simulation and learning?
Can we, will we, and should we close the sim2real gap?
What makes a good differentiable simulator?
How can we ensure the quality and longevity of open-source simulators?
+ questions from the audience
Lunch break 12:00 –13:45
Applications in policy learning and control (1/2)
Applications in policy learning and control (1/2)
13:45
Invited talk
Invited talk
Hae-Won Park
Contact-Implicit Model Predictive Control on Legged Robots
Contact-Implicit Model Predictive Control on Legged Robots
14:15
Invited talk
Invited talk
Yuval Tassa
Finite differencing is all you need: lessons and insights
Finite differencing is all you need: lessons and insights
14:45
Invited talk
Invited talk
Lin Shao
Large-Scale Manipulation Data Acquisition through Differentiable Simulation and Rendering
Poster session + coffee break 15:15 - 16:00
Applications in policy learning and control (2/2)
Applications in policy learning and control (2/2)
16:00
Invited talk
Invited talk
Emo Todorov
Leveraging the internals of MuJoCo physics to improve model-based control
Improving simulation from data (2/2)
Improving simulation from data (2/2)
16:30
Invited talk (remote)
Invited talk (remote)
Kelsey Allen
Simulation from states and sensors
17:00 Panel discussion II
What kind of gradients do we need, if any?
Is training in simulation the right approach for obtaining safe and performant policies?
How can we (and should we) reduce the ecological footprint of policy training?
+ questions from the audience
17:30 Closing remarks
17:40 End of workshop
18:00 Social event sponsored by Nvidia
Page updated
Report abuse