Our solution provided 100+ years of simulated driving experience, accelerating Volkswagen’s autonomous driving R&D.
Meet our client
Client:
Industry:
Market:
Technology:
Client’s Challenge
Volkswagen needed to speed up autonomous driving development by simulating real-world scenarios. The goal was to apply the Sim2Real paradigm to reduce reliance on limited real-world data while ensuring accurate transfer of models to physical vehicles.
Our Solution
We developed a reinforcement learning (RL) neural network trained in a simulated environment. To bridge the sim-to-real gap, we built a data collection pipeline that carefully controlled variables like weather, traffic density, and driving trajectories. This ensured smooth transfer from simulation to real-world driving. The project’s success was showcased at the ICRA conference, highlighting its innovative approach.
Client’s Benefits
Our solution provided 100+ years of simulated driving experience, accelerating Volkswagen’s autonomous driving R&D. The project concluded with a successful real-world test drive at Volkswagen’s HQ in Wolfsburg, showcasing the system’s effectiveness in real-world conditions.
