Sim2Real
A technical direction concerned with transferring models trained in simulated environments to real-world scenarios. The central challenge is bridging the 'reality gap' between simulation and reality. The approach is widely applied in robotics and physics-based modeling.
Timeline (last 90 days)
The RealPDE competition features two core tracks: Sim2Real and LTTTA.
The core challenge of the Sim2Real track is to generalize neural operator models trained on CFD data to real PIV data.
In 2019, OpenAI demonstrated that a dexterous hand policy trained in simulation was successfully transferred to a real Shadow Hand to complete a Rubik's Cube solve.
All Facts (3)
The RealPDE competition features two core tracks: Sim2Real and LTTTA.
50%UnverifiedThe core challenge of the Sim2Real track is to generalize neural operator models trained on CFD data to real PIV data.
50%UnverifiedIn 2019, OpenAI demonstrated that a dexterous hand policy trained in simulation was successfully transferred to a real Shadow Hand to complete a Rubik's Cube solve.
50%