Predicting the Unpredictable: The Scene Transformer and the Future of Safe Autonomy

We break down Scene Transformer, a scene-centric, attention-driven model that jointly predicts trajectories for cars, pedestrians, and cyclists. Learn why moving from marginal to joint prediction matters, how a global view improves accuracy, and what a 15% leap on real-world data could mean for safer, smarter self-driving cars—and the broader future of predictive motion.

Note: This podcast was AI-generated, and sometimes AI can make mistakes. Please double-check any critical information.

Sponsored by Embersilk LLC