[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85991-en":3,"doc-seo-85991-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},85991,1374391975076,"Riley","https://ap-avatar.wpscdn.com/avatar/14000253ca4ec9f6853?x-image-process=image/resize,m_fixed,w_180,h_180&k=1783305029341752051",8,"Research & Report","Dual-Process Atomic Skill Learning Decoupling Semantic Reasoning and Real-Time Control","Language-conditioned imitation learning enables robots to follow natural-language instructions, yet multi-step compositional tasks remain difficult to generalize. Existing hierarchical decompositions often couple slow skill-level reasoning with fast action generation, causing training instability and skill codebook collapse. Dual-Process Atomic Skill Learning (DASL) introduces asynchronous decoupling between interpretable discrete skills and real-time motion control. A slow semantic policy predicts VQ-quantized skills, while a high-frequency policy uses latent diffusion with a decision transformer for accurate actions, improving stability and compositional generalization.","arXiv :2607 . 10625v1 [ cs .RO] 12 Jul 2026  \nDual-Process Atomic Skill Learning: Decoupling Semantic Reasoning and Real-Time Control  \nJun Chen 1∗, Erdent Bao2∗, Wenlong Dong3 , Jierui Liu4 , Qi Cai4 , Hao Wan 1 , Shaopeng Li5 , Weijun Qin5 , Jing Liang 1 , Huiping Zhuang4†  \n1University of Electronic Science and Technology of China 2Huazhong University of Science and Technology  \n3 Southern University of Science and Technology 4 South China University of Technology 5EbTech Co. Ltd.  \n∗ Equal contribution. †Corresponding author. Email: [hpzhuang@scut.edu.cn](hpzhuang@scut.edu.cn)  \nAbstract  \nLanguage-conditioned Imitation Learning (IL) is essential for enabling robots to perform complex tasks following natural language instructions. However, generalizing to multi-step compositional tasks remains a significant challenge. While hierarchical approaches attempt to address this by decomposing tasks into atomic skills, existing methods often suffer from training instability and codebook collapse due to the tight coupling between high-level skill reasoning and low-level action generation in joint training paradigms. Inspired by the Dual-Process Theory of cognition, we propose Dual-Process Atomic Skill Learning (DASL), a novel asynchronous hierarchical imitation learning framework that decouples slow semantic reasoning from fast, real-time motion control. DASL comprises a Slow-Frequency Policy that predicts interpretable, discrete skills via Vector Quantization, and a High-Frequency Policy that leverages a latent diffusion model and a Decision Transformer to generate precise actions conditioned on these latent skills. By asynchronously coordinating these modules and utilizing diffusion to structure the latent space, our framework mitigates the skill codebook interference problem common in joint training paradigms. Evaluations across simulation benchmarksand experiment demonstrate that DASL significantly outperforms state-of-the-art baselines, excelling in skill acquisition and compositional generalization to unseen instructions. GitHub page: [https://github.com/Hatakekaka/DASL](https://github.com/Hatakekaka/DASL)  \n1 Introduction  \nLanguage-conditioned imitation learning (IL) aims to train robots to execute tasks specified by natural language instructions using paired language-trajectory demonstrations. However, this objective becomes increasingly arduous when tasks involve sequential execution of multiple sub-goals [10, 38] . A key strategy to address this complexity exploits the inherent hierarchical structure of natural language to transform complex task solving into learning atomic skills. Recombining these atomic skills enables robots to generalize to unseen task configurations, a capability particularly critical in data-constrained regimes where covering the entire task space is infeasible.  \nDespite progress in hierarchical skill learning methods [19, 26], a significant limitation persists: highlevel skill transition and low-level action generation  \nPreprint.  \n(a) Baselines (b) Ours  \nFigure 1: Baselines vs. our dual-process control.(a) Baselines: synchronous high-frequency skill and action generation. (b) Ours: asynchronous decoupling, with slow-frequency skill reasoning and high-frequency action execution.  \nremain tightly coupled. This necessitates end-to-end co-training of both policies, inducing severe instability during skill learning. The latent low-level and language-conditioned high-level policies become intertwined, causing mutual interference. This frequently leads to codebook collapse, forcing reliance on cumbersome multi-stage training strategies or additional constraints to optimize and stabilize the skill codebook [8, 17, 13] .  \nInspired by Kahneman’s Dual-Process Theory, distinguishing fast, intuitive \"System 1\" from slow, deliberative \"System 2\" cognition [27, 37, 4], we argue similar decoupling benefits language-conditioned skill learning. While recent Vision-Language-Action (VLA) models explore dual-syst","cbCailR5lujkflkb","https://ap.wps.com/l/cbCailR5lujkflkb","pdf",20265675,6,1,28,"English","en",105,"# Introduction\n## Problem: coupling in hierarchical imitation learning\n## Dual-process motivation and decoupling hypothesis\n## Proposed method: DASL framework\n## Experimental evaluation and contributions","[{\"question\":\"What problem does DASL target in language-conditioned hierarchical imitation learning?\",\"answer\":\"DASL targets the tight coupling between high-level skill reasoning and low-level action generation that leads to training instability and codebook collapse in joint training paradigms.\"},{\"question\":\"How does DASL decouple semantic reasoning from real-time control?\",\"answer\":\"DASL uses an asynchronous hierarchical setup with a Slow-Frequency Policy for interpretable discrete skills and a High-Frequency Policy for precise control conditioned on the selected skills.\"},{\"question\":\"What components does DASL use to generate skills and actions?\",\"answer\":\"The Slow-Frequency Policy predicts skills and discretizes them using Vector Quantization, while the High-Frequency Policy trains with a skill-conditioned latent diffusion model and uses a Decision Transformer to generate real-time actions at inference.\"}]",1784207628,71,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"dual-process-atomic-skill-learning-decoupling-semantic-reasoning-and-real-time-control","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/dual-process-atomic-skill-learning-decoupling-semantic-reasoning-and-real-time-control/85991/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-07-26","2026-07-16",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What problem does DASL target in language-conditioned hierarchical imitation learning?","Question",{"text":76,"@type":77},"DASL targets the tight coupling between high-level skill reasoning and low-level action generation that leads to training instability and codebook collapse in joint training paradigms.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does DASL decouple semantic reasoning from real-time control?",{"text":81,"@type":77},"DASL uses an asynchronous hierarchical setup with a Slow-Frequency Policy for interpretable discrete skills and a High-Frequency Policy for precise control conditioned on the selected skills.",{"name":83,"@type":74,"acceptedAnswer":84},"What components does DASL use to generate skills and actions?",{"text":85,"@type":77},"The Slow-Frequency Policy predicts skills and discretizes them using Vector Quantization, while the High-Frequency Policy trains with a skill-conditioned latent diffusion model and uses a Decision Transformer to generate real-time actions at inference.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":107,"slug":138},19,"General","general"]