摘要
“人工鱼”利用人工生命方法创作计算机动画 .为了进一步研究动画角色的自进化特性 ,本文提出并建立了一种基于时间差分方法TD的BP神经网络 -TDBP ,人工鱼依靠TDBP网络可以进行简单的识别和在环境中自动的强化学习 .文中还研究了人工鱼的感知范围和遮挡等问题 ,利用模式识别等技术设计了人工鱼的感知系统 .实验表明 ,基于自学习的“人工鱼”
In artificial fish, computer animation is created with artificial life. To study Self-Evolutionary of animation role, a BP neural network based on Time Difference (TD)-TDBP is presented, and artificial fish can automatically learn in environment. Perceptive area and visibility of artificial fish are studied, and perception system is designed using the technologies of pattern recognition. The experimental results have shown that perception system of artificial fish based on self-learning provides artificial life with more independence and intelligence.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2004年第12期2041-2045,共5页
Acta Electronica Sinica
基金
国家自然科学基金 (No 60 0 750 1 2
60 3750 38)
清华大学智能技术与系统国家重点实验室开放课题 (No 0 1 0 9)
关键词
人工生命
自学习
强化学习
人工神经网络
Animation
Computer simulation
Learning systems
Mathematical models
Neural networks
Pattern recognition
Sensory perception
Visibility
