Research map / Probabilistic and scientific modeling / Dynamical systems
CLaD: Planning with Grounded Foresight via Cross-Modal Latent Dynamics
Andrew Jeong, Jaemin Kim, Sebin Lee, Sung-Eui Yoon
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add semantic foresight conditioning to the diffusion policy (ablation) | CLaD diffusion policy · LIBERO-LONG | average success rate (%): 84.8 -> 91.5 | Improved |
| Condition the policy on proprioceptive foresight only (ablation) | CLaD diffusion policy · LIBERO-LONG | average success rate (%): 84.8 -> 50.4 | Hurt |
| Add proprioceptive foresight alongside semantic foresight (full cross-modal foresight) (ablation) | CLaD · LIBERO-LONG | average success rate (%): 91.5 -> 94.7 | Improved |
| Add auxiliary reconstruction loss Lrecon (ablation) | CLaD stage 1 · LIBERO-LONG | average success rate (%): 86.1 -> 94.7 | Improved |
| Use proprioceptive-queries-semantic cross-attention instead of symmetric self-attention (ablation) | CLaD dynamics model · LIBERO-LONG | average success rate (%): 86.7 -> 94.7 | Improved |
| Use proprioceptive queries over semantic instead of the reverse direction (ablation) | CLaD dynamics model · LIBERO-LONG | average success rate (%): 93.8 -> 94.7 | Improved |
| Use CLaD instead of LBP latent planner | latent planning · LIBERO-LONG Task 9 | success rate (%): 60.0 -> 81.7 | Improved |
| Use CLaD instead of LBP regarding planning latency and size | latent planning · LIBERO-LONG | planning time (s): 0.008 -> 0.012 | Hurt |
Closest papers by meaning
- Beyond Dense Futures: World Models as Structured Planners for Robotic Manipulation (ECCV 2026 · similarity 0.90)
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.87)
- Generate Subgoal Images before Act: Unlocking the Chain-of-Thought Reasoning in Diffusion Model for Robot Manipulation with Multimodal Prompts (CVPR 2024 · similarity 0.86)
- InfoCon: Concept Discovery with Generative and Discriminative Informativeness (ICLR 2024 · similarity 0.86)
- Chain-of-Action: Trajectory Autoregressive Modeling for Robotic Manipulation (NeurIPS 2025 · similarity 0.86)
- $\textit{HiMaCon:}$ Discovering Hierarchical Manipulation Concepts from Unlabeled Multi-Modal Data (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
