| Session Name: | Reinforcement Learning for Efficient Cars and Tracks Design in Racing Games |
| Speaker(s): | Minggao Wei |
| Company Name(s): | NetEase Games AI Lab |
| Track / Format: | Programming |
Did you know free users get access to 30% of content from the last 2 years?
Get your team full access to the most up to date GDC content
| Overview: | Traditionally in a racing game, it is time-consuming for the game designers to manually verify and adjust the performance of cars on every track. This presentation introduces an AI-driven approach based on deep reinforcement learning to free the designers from boring repetitions. With this technique, the optimal racing trace for all cars can be figured out in less than an hour, making it possible to do a thorough evaluation. Besides that, the drift area, which is highly related to the track difficulty, can be visualized to help the assessment. Even without the game designers' prior knowledge, this AI system considerably reduces the game development lifetime and provides a more accurate and balanced analysis. |
Recommended
