Lessons from Unifying Simulation and RL Training for Autonomous Trucking

Autonomous trucking is hard. Torc generalizes the task planning stack with reinforcement learning, running two very different RL loops, object simulation and end-to-end with neural rendering, both on Ray.

Neil Wadhvana and Yashovardhan Chaturvedi from Torc Robotics share how they adapted existing LLM post-training frameworks for the end-to-end loop and found the two scaling problems surprisingly alike.

You'll leave with a blueprint for scaling RL for autonomy.