← Back to all news
arXiv AI By Changchuan Yang, Haoxuan Xu, Wenbo Chen, Shuai Ren, Jianlong Zheng, Huarui Zhang, Tianfu Li, Guanzhong Tian

PMTRM: Pseudo-Memory Temporal Re-encoding Module for Embodied Policy Learning

Read the original on arXiv AI →

The paper introduces PMTRM, a lightweight plug‑in module that encodes a bounded history of executed states and actions into a latent sequence to help policies disambiguate repeated motions in robotic manipulation. It uses a temporal heterogeneity objective to penalize similarity between distant positions, along with anchor and reconstruction losses to preserve action‑relevant information. The module is trained progressively on synthetic and real robot data, then jointly with existing policies, and experiments show improved task success on phase‑ambiguous tasks with minimal extra computation.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv AI
Jul 24

VPWEM: Non-Markovian Visuomotor Policy with Working and Episodic Memory

arXiv:2603. 04910v2 Announce Type: replace-cross Abstract: Imitation learning from human demonstrations has achieved significant success in robotic control, yet most visuomotor policies still condition on single-step observations or short-context histories, making them struggle with non-Markovian tasks that require long-term memory.

By Yuheng Lei, Zhixuan Liang, Hongyuan Zhang, Ping Luo
llmsragdiffusionroboticsmultimodalbenchmarks
More like this →
arXiv Machine Learning
Sep 14

Dynin-Robotics: Omnimodal Unified Diffusion Vision-Language-Action Model

Dynin‑Robotics introduces an omnimodal masked‑diffusion backbone, Dynin‑Omni, that jointly represents language, visual observations, goals, and actions as discrete tokens. By conditioning on different spans, the same model learns action prediction, next‑observation prediction, goal‑state prediction, and trajectory‑to‑instruction reconstruction, enabling test‑time scaling through goal prediction and action‑candidate evaluation. The system, pretrained on 1.33 million trajectories from 48 Open X‑Embodiment datasets, achieves competitive performance on LIBERO, zero‑shot LIBERO‑Plus, and a 78.4 % success rate on a Franka Research 3 robot, while a block‑parallel implementation speeds up action decoding by up to 29.2×.

By Hoeun Lee, Jaeik Kim, Jusang Oh, Jinhyeok Kim, Geon Choi, Hyeonggeun Kim, Jaeyoung Do
diffusionroboticsmultimodal
More like this →