Research map / Image generation / Autoregressive and latent generation

Diffusion Models for Robotics

AAAI 2025 · Image generation › Autoregressive and latent generation · approach: Diffusion planning and policies · 2 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

A short AAAI 2025 proposal outlining planned diffusion-model work for robotics: colored-noise trajectory generation, Transfer Entropy-regularized 3D generation, and PID-based interpretable scene generation, with no experiments reported.

Problem

Diffusion models for robotics must handle uncertainty robustly in trajectory generation, 3D image generation, and scene understanding, but existing DMs use white Gaussian noise and lack directional information-flow and interpretability controls.

Approach

Replace white Gaussian noise with colored Gaussian noise for trajectory diffusion, add a Transfer Entropy term to the 3D voxel-editing loss to strengthen text-to-visual information flow, and use Partial Information Decomposition to maximize redundant and minimize synergistic information when training the Stable Diffusion UNet.

Key results

• No experimental results are reported; the paper describes planned evaluation of trajectory quality, robustness, sim-to-real transfer, and interpretability.

Limitations

• No experiments or quantitative results are presented; claims are prospective.
• Claim that colored noise improves DDPM performance is stated as an observation without evidence in the text.
• Computational demands of training and deploying diffusion models are acknowledged as a challenge.

Experimental setup

Metrics: path length curvature energy consumption number of collisions

Baselines: deterministic approaches reinforcement learning-based methods

Builds on: Denoising Diffusion Probabilistic Models Diffuser text-guided voxel editing Partial Information Decomposition Stable Diffusion

Closest papers by meaning

• READ: Retrieval-Enhanced Asymmetric Diffusion for Motion Planning (CVPR 2024 · similarity 0.88)
• Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024 · similarity 0.87)
• Prediction with Action: Visual Policy Learning via Joint Denoising Process (NeurIPS 2024 · similarity 0.87)
• Tracing the Roots: Leveraging Temporal Dynamics in Diffusion Trajectories for Origin Attribution (NeurIPS 2025 · similarity 0.87)
• Multi-Robot Motion Planning with Diffusion Models (ICLR 2025 · similarity 0.87)
• Model-based Diffusion for Trajectory Optimization (NeurIPS 2024 · similarity 0.87)