arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2102.09812cs.LGcs.AIcs.RO
收藏

深度潜在竞争:在潜在空间中使用视觉控制策略学习比赛

Deep Latent Competition: Learning to Race Using Visual Control Policies in Latent Space

MIT CSAIL(麻省理工学院计算机科学与人工智能实验室);MIT LIDS(麻省理工学院信息与决策系统实验室)

Wilko Schwarting, Tim Seyde, Igor Gilitschenski, Lucas Liebenwein, Ryan Sander, Sertac Karaman, Daniela Rus

摘要:

在多智能体环境中学习竞争行为,比如赛车,需要对潜在的对抗性互动进行长期推理。本文提出了深度潜在竞争(DLC),一种新的强化学习算法,学习竞争的视觉控制策略,通过自我发挥的想象力。DLC代理想象多智能体交互序列的学习世界模型,结合了联合过渡函数与对手的观点预测的紧凑的潜在空间。想象中的自我游戏减少了现实世界中昂贵的样本生成,而潜在的表征使规划能够随着观察维度优雅地扩展。我们证明了我们的算法在学习竞争行为的有效性,一个新的多智能体赛车基准,需要规划从图像观察。代码和视频可在https://sites.google.com/view/deep-latent-competition获得。

英文摘要:

Learning competitive behaviors in multi-agent settings such as racing requires long-term reasoning about potential adversarial interactions. This paper presents Deep Latent Competition (DLC), a novel reinforcement learning algorithm that learns competitive visual control policies through self-play in imagination. The DLC agent imagines multi-agent interaction sequences in the compact latent space of a learned world model that combines a joint transition function with opponent viewpoint prediction. Imagined self-play reduces costly sample generation in the real world, while the latent representation enables planning to scale gracefully with observation dimensionality. We demonstrate the effectiveness of our algorithm in learning competitive behaviors on a novel multi-agent racing benchmark that requires planning from image observations. Code and videos available at https://sites.google.com/view/deep-latent-competition.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑