Breakthroughs in Gaming & Agents
See how we're accelerating the future of gaming and interactive agents by bringing the latest research in competitive play and human-centric technology to life.
Training Reinforcement Learning Agents for Video Games
Reinforcement learning research involves training agents to master the precise controls and creative solutions required by modern video games. These complex environments inspire innovations that apply to diverse real-world domains beyond gaming.
Interacting Seamlessly with Humans and Other Agents
Multi-agent learning research focuses on creating robust, tunable agents that interact seamlessly in environments with informal rules. These techniques are essential for navigating complex behaviors in both virtual and physical worlds.
Creating an Ecosystem for Game-AI Training
Platform engineering provides a scalable ecosystem for the rapid deployment of learning algorithms at scale. This infrastructure streamlines the training process, making reinforcement learning a more trustworthy and repeatable pursuit for production-grade domains.
Industry Accolades & Awards
Gran Turismo
A Super-Human Vision-Based Reinforcement Learning Agent for Autonomous Racing in Gran Turismo
BeTAIL
Behavior Transformer Adversarial Imitation Learning from Human Racing Gameplay
Autonomous Racing
Real-Time Trajectory Generation via Dynamic Movement Primitives for Autonomous Racing
Blog
Sony AI’s Deep RL Team on Why the Hardest Problems Still Matter
