Extended Machine Learning
summary
A photorealistic XR digital twin of a parking lot is built in Unity and driven by Unity ML‑Agents with a Python/TensorFlow backend and C# scripting; training runs with 16 simultaneous agents and a network of 512 nodes per layer across four hidden layers. Rewards are +0.8/+0.2 and penalty −0.01; after about 5 million cycles the learned policy transfers as saved weights and cuts required actions from 200→40, working reliably in simulation.
AI-written summary · about
chapters (8)
- 0:00Introduction to speakers
- 1:17XR technology overview
- 10:27XR industry impact and adoption
- 22:30Machine learning basics
- 31:31Digital twins explained
- 36:54Reinforcement learning for parking
- 47:36Unity simulation demo
- 57:02Training results and next steps
related talks
30mBuilding, designing, teaching and training simulation environments for Machine Learning
43mBuilding Environments for Autonomous Vehicle Training
36mImplementing Machine Learning the Unity way
10mMachine Learning & Unity ML-Agents update
48mMachine Takeover - Improving Your Game With Machine Learning & AI
41mAutonomous vehicle development with Unity
47mArtificial Intelligence and XR
35mCrowd dynamics, RL and Unity: A Journey