Newest papers

De-Diffusing WAMs: Action-Sufficient Direct Prediction for Robot Control

Shaoyi Zheng, Zhongtao Jiang, Yue Peng, Wenbo Lu, Shunchi Zhang, Xi Wang, Shengji Wan · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28122 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Model aksi dunia kaskade (WAM) ini menggunakan masa depan visual yang diprediksi untuk mempengaruhi tindakan robot, tetapi mewarisi kesimpulan iteratif yang mahal dari para ahli yang dilatih dalam difusi. Kami bertanya apa yang harus dipertahankan oleh masa depan yang diprediksi satu langkah untuk mendukung prediksi tindakan satu langkah yang akurat. Dalam WAM yang dinilai, sampel ulang suara video memiliki efek sedikit pada tindakan di bawah, namun naif truncating kedua aliran ke satu langkah secara signifikan mengurangi kesuksesan tugas. Pengawasan ruang video hanya sebagian menutup kesenjangan ini, dan analisis sensitivitas lokal menunjukkan bahwa kesalahan prediksi visi memiliki konsekuensi yang sangat tidak seimbang pada akurasi tindakan. Temuan ini memotivasi DirectWAM, yang mengkonversi video pra-pelatihan dan para ahli aksi menjadi prediktor langkah tunggal tanpa pengambilan sampel stochastic

Machine translation (Meta M2M100); the English original follows.

Abstract

Cascade world-action models (WAMs) use predicted visual futures to condition robot actions, but inherit costly iterative inference from diffusion-pretrained experts. We ask what a single-step predicted future must preserve to support accurate single-step action prediction. In the evaluated WAM, resampling video noise has little effect on downstream actions, yet naively truncating both streams to one step substantially reduces task success. Video-space supervision only partially closes this gap, and a local sensitivity analysis shows that vision prediction errors have highly unequal consequences for action accuracy. These findings motivate DirectWAM, which converts pretrained video and action experts into single-step predictors without stochastic sampling through action-sufficient distillation, while retaining the explicit future-to-action cascade. After learning a one-step action model, we train the future predictor by decoding actions from its own predicted futures and backpropagating demonstration action error through the expert, with an optional future-latent reconstruction loss. The resulting objective evaluates the student-generated future through its downstream control consequences rather than relying on representation matching alone. DirectWAM achieves 97.1% average success across four LIBERO suites and 85.4% on RoboTwin 2.0, outperforming Flash-WAM at the same one-evaluation-per-stream budget. This advantage has also extended to three real-world tasks that we designed. Compared with the full iterative teacher, it reduces inference latency by $28.0\times$.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Artificial intelligence, Inference, Machine learning, Task (project management), Action (physics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange