Communications of the ACM
How a Computer Learns to Dribble: Practice, Practice, Practice
Researchers at Carnegie Mellon University and DeepMotion have developed a physics-based, real-time method for controlling animated characters that can learn basketball dribbling skills from experience.
Credit: Carnegie Mellon University
Carnegie Mellon University and DeepMotion researchers have developed a physics-based, real-time method for controlling animated characters that can learn basketball dribbling skills from experience.
The system learns from motion capture of movements performed by people dribbling basketballs.
The researchers used deep reinforcement learning to enable the model to identify important details associated with dribbling.
The system learned the skills in two stages: it first mastered locomotion, and then learned how to control animated characters' arms and hands to manipulate the ball.
DeepMotion's Libin Liu says, "The technology can be applied beyond sport simulation to create more interactive characters for gaming, animation, motion analysis, and in the future, robotics."
The researchers are presenting the work this week at ACM SIGGRAPH 2018 in Vancouver, British Columbia.
From Carnegie Mellon University
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found