基于世界模型的潜在能量动作规划
Original title: Latent Energy Action Planning with World Models
ResearchAI 70
Source: arXiv cs.LGPublish time unverified
arXiv:2609.03294v1 Announce Type: new Abstract: Latent world models support efficient model predictive control from high-dimensional observations, yet optimizing a single learned latent objective can favor action sequences whose decoder-predicted terminal descriptor does not match the goal descriptor. We introduce Latent Energy Action Planning (LEAP), which treats the complete action horizon as a differentiable variable and optimizes it through a frozen LeWorldModel (LeWM).