[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86085-en":3,"doc-seo-86085-105":29,"detail-sidebar-cat-0-en-105":91},{"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":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},86085,2336464648746,"Skyler","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",6,"Technology","Edge Physical AI Deployment of Vision Transformers on Heterogeneous Edge GPU Targeting Autonomous Vehicles","Physical AI systems such as autonomous vehicles rely on transformer-based perception that must meet strict edge latency and energy budgets. Yet heterogeneous edge-GPU deployment is constrained by underutilized hardware engines and operators incompatible with accelerators, reducing throughput per watt. The paper introduces Heterogeneous Frame Dispatch Scheduling (H-FraDS), a hardware-aware frame scheduling method for transformer inference on a modern NVIDIA edge GPU. It dispatches frames across GPU and dual DLA cores with fixed ratios while adapting incompatible components for DLA execution and using an optical-flow accelerator for estimation, achieving real-time performance.","Edge Physical AI Deployment of Vision Transformers on Heterogeneous Edge GPU Targeting  \nAutonomous Vehicles  \nAshiyana Abdul Majeed, Member, IEEE, Mahmoud Meribout, Senior Member, IEEE, Neethu Joseph, Abel Kidane Haile, and Mohammad Abdullah Al Faruque, Senior Member, IEEE  \narXiv :2607 . 10942v1 [ cs .AR] 12 Jul 2026  \nAbstract—Physical AI systems, such as autonomous vehicles and intelligent machines, require transformer-based perception models that satisfy stringent edge latency and energy constraints. However, heterogeneous edge-GPU deployment remains limited by underutilized hardware engines and accelerator-incompatible operators, causing fragmented execution and lower throughput per watt. This paper presents Heterogeneous Frame Dispatch Scheduling (H-FraDS), a hardware-aware frame scheduling methodology for transformer inference on a recent NVIDIA edge GPU. H-FraDS routes frames across the GPU and dual deep learning accelerator (DLA) cores using fixed dispatch ratios to improve utilization under latency and power constraints. To enable scheduling, incompatible transformer components are adapted for DLA execution by reshaping tensors, approximating error function (ERF) with tanh, and replacing layer normalization with bounded tanh. The adapted model maintains a 92% F1 score, with only a 2% reduction from the original. Optical flow accelerator (OFA) is further used for inference-side optical-flow estimation. To the best of the authors’ knowledge, prior work has not addressed these combined issues. Using Swin Transformer for autonomous-driving perception, H-FraDS Balanced Dispatch (1:2) achieves 125.93 FPS, a 2.36× speedup over standalone adaptedDLA execution, 4.0 FPS/W, and ≈24 ms DLA latency, satisfying 30 FPS real-time operation; the GPU-DLA-OFA case achieves a 2.02× DLA throughput speedup.  \nIndex Terms—deep learning accelerator (DLA), transformer networks, Heterogeneous Frame Dispatch Scheduling (H-FraDS), heterogeneous computing, optical flow accelerator (OFA), autonomous driving, embedded vision systems, edge AI  \nI. INTRODUCTION  \nModern edge-AI systems must execute real-time perception and signal-processing workloads under strict latency, power, and thermal constraints. Applications such as autonomous driving, mobile robotics, intelligent surveillance, industrial inspection, and embedded vision increasingly rely on compact heterogeneous platforms that combine GPU resources with fixed-function engines, including deep learning accelerators (DLAs), vision accelerators, video engines, and optical-flow engines. Their practical performance, therefore, depends not only on model accuracy but also on whether the deployment pipeline can exploit the available hardware engines concurrently  \nAshiyana Abdul Majeed, Dr. Mahmoud Meribout, Neethu Joseph, and Abel Kidane Haile are with the Department of Computer and Information Engineering, Khalifa University, Abu Dhabi, UAE (email: [100059454@ku.ac.ae](100059454@ku.ac.ae), [mahmoud.meribout@ku.ac.ae](mahmoud.meribout@ku.ac.ae), [100069410@ku.ac.ae](100069410@ku.ac.ae),  \n[ku100053692@alumni.ku.ac.ae](ku100053692@alumni.ku.ac.ae)). Dr. Mohammad Abdullah Al Faruque is with the Department of Electrical Engineering and Computer Science at the University of California, Irvine, CA 92697, USA (email: [alfaruqu@uci.edu](alfaruqu@uci.edu)).  \n[1] . Autonomous driving is a representative high-pressure case: perception must support detection and localization under varied environmental conditions within tight computational budgets, since latency or accuracy failures can propagate to prediction, planning, and control. This creates a central trade-off among accuracy, latency, and efficiency, particularly as highperforming models become more computationally demanding. This work is motivated by the transition from conventional edge AI to edge physical AI, where perception models are no longer isolated classifiers but foundational components inside closed-loop physical agents. In physical AI syste","cbCaik4QDjwNbjyN","https://ap.wps.com/l/cbCaik4QDjwNbjyN","pdf",3799782,1,14,"English","en",105,"# Abstract\n# Introduction","[{\"question\":\"What problem does H-FraDS address in heterogeneous edge GPU deployment?\",\"answer\":\"It targets limited utilization caused by underused hardware engines and accelerator-incompatible operators, which fragment execution and reduce throughput per watt.\"},{\"question\":\"How does H-FraDS improve transformer inference utilization and throughput?\",\"answer\":\"It routes frames across the GPU and two DLA cores using fixed dispatch ratios under latency and power constraints, improving engine utilization for transformer inference.\"},{\"question\":\"How is the transformer model adapted to run on the DLA cores?\",\"answer\":\"It reshapes tensors for DLA execution, approximates the error function (ERF) using tanh, and replaces layer normalization with a bounded tanh variant to match DLA capabilities.\"}]",1784208407,35,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"edge-physical-ai-deployment-of-vision-transformers-on-heterogeneous-edge-gpu-targeting-autonomous-vehicles","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/edge-physical-ai-deployment-of-vision-transformers-on-heterogeneous-edge-gpu-targeting-autonomous-vehicles/86085/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does H-FraDS address in heterogeneous edge GPU deployment?","Question",{"text":75,"@type":76},"It targets limited utilization caused by underused hardware engines and accelerator-incompatible operators, which fragment execution and reduce throughput per watt.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does H-FraDS improve transformer inference utilization and throughput?",{"text":80,"@type":76},"It routes frames across the GPU and two DLA cores using fixed dispatch ratios under latency and power constraints, improving engine utilization for transformer inference.",{"name":82,"@type":73,"acceptedAnswer":83},"How is the transformer model adapted to run on the DLA cores?",{"text":84,"@type":76},"It reshapes tensors for DLA execution, approximates the error function (ERF) using tanh, and replaces layer normalization with a bounded tanh variant to match DLA capabilities.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":45,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":45,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]