[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84895-en":3,"doc-seo-84895-105":30,"detail-sidebar-cat-0-en-105":83},{"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},84895,8796095461610,"Oliver","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Calf-Integrated Arms for Bimanual Quadruped Loco-Manipulation","Quadruped loco-manipulation often forces a trade-off between manipulation ability and stable stance, limiting ground-level reach, true two-handed work, and long-horizon action sequencing. The paper integrates a manipulator—prismatic slider, two revolute joints, and a gripper—into each front calf of a Unitree Go2. Two arms grasp objects at ground level and execute bimanual manipulation with all four feet planted, while the base can still walk when carrying a single arm. A vision-language model selects skills from a predefined library at each skill boundary for long-horizon autonomy, validated in simulation on cabinet, cooperative lift, and inter-arm handover tasks.","Calf-Integrated Arms for Bimanual Quadruped Loco-Manipulation  \nYan Pan 1 , Yuanchuan Ren 1 , Chipui Chan 1 , Jingcheng Sun 1 and Chengxu Zhou 1  \narXiv :2607 .06186v2 [ cs .RO] 11 Jul 2026  \nAbstract—Most quadruped loco-manipulation designs trade manipulation capability against stance. A trunk-mounted armsits high and usually carries a single arm; using the legs as manipulators lifts the manipulating leg off the ground; and even leg-mounted grippers reach two-handed tasks only by rearing onto the hind legs. This paper integrates a manipulator with a prismatic slider, two revolute joints, and a gripper into each front calf of a Unitree Go2. The two arms grasp objects at ground level and manipulate with both hands while all four feet stay planted, without rearing. With one arm carrying, the base stays free to walk. A vision-language model sequences skills from a predefined library at each skill boundary, conditioned on the head-camera image and task state, for long-horizon autonomy. In simulation, the design performs three bimanual tasks: a long-horizon cabinet task under autonomous skillselection, a cooperative two-handed lift, and an inter-arm handover. Index Terms—uadruped Robots, Loco-Manipulation, Morphological Design, Integrated Gripperuadruped Robots, Loco-Manipulation, Morphological Design, Integrated GripperQ  \nI. INTRODUCTION  \nQuadruped robots move capably through cluttered, humanscale environments [1], yet endowing them with manipulation, the ability to act on the objects around them, remains an open design problem. Loco-manipulation, the combination of locomotion and manipulation, would extend them to tasks such as search and rescue, inspection, and logistics. Consider clearing a cluttered room: the robot must pick litter off the floor, carry a basket in one hand while opening a cabinet with the other, and choose the order of these actions itself. This calls for three capabilities at once: reaching objects at ground level, using both hands together, and sequencing skills over a long horizon.  \nExisting loco-manipulation designs rarely offer all three (Section II) . A trunk-mounted arm reaches objects but sits high on the body and usually carries a single arm. Using the bare legs as manipulators gives up the support stance, and leg-mounted grippers free both hands only by rearing onto the hind legs, which pins the base in place. No design keeps all four feet planted while manipulating with two hands down to ground level, nor pairs this with a planner that selects skills from the current scene and task state. The integrated leg-arm design proposed here provides all three capabilities: a manipulator built into each front calf of a Unitree Go2, so the quadruped grasps objects at ground level with both arms.  \nThis paper makes two contributions:  \nThis work was partially supported by the Advanced Research and Invention Agency [grant number SMRB-SE01-P06] .  \n1Dept. of Computer Science, University College London, UK. [chengxu.zhou@ucl.ac.uk](chengxu.zhou@ucl.ac.uk)  \n1) Integrating a manipulator with a prismatic slider, two revolute joints, and a gripper into each front calf of a Unitree Go2, enabling bimanual loco-manipulation and ground-level grasping.  \n2) Introducing a vision-language model that selects the next skill from a predefined library at each skill boundary, conditioned on the head-camera image and task state, demonstrated on a long-horizon bimanual task in simulation.  \nThe paper details the mechanical design (Section III) and the control framework (Section IV), reports the simulation study including the observed failure modes (Section V), and concludes in Section VI.  \nII. RELATED WORK  \nLoco-manipulation designs differ in where the manipulator is placed. The most common approach mounts a dedicated arm on the trunk: articulated arms on ANYmal [2] and Spot [3] perform dynamic grasping, model-predictive and whole-body controllers coordinate the arm with the base [4],[5], and learning-based controllers extend","cbCaib4LD06uzdvG","https://ap.wps.com/l/cbCaib4LD06uzdvG","pdf",6545273,2,1,6,"English","en",105,"# Introduction\n# Related Work","[{\"question\":\"How is long-horizon skill sequencing achieved for autonomy?\",\"answer\":\"A vision-language model sequences skills from a predefined library at each skill boundary, conditioned on the head-camera image and the task state, enabling long-horizon autonomy in simulation.\"}]",1784199158,15,{"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":78,"head_meta":80,"extra_data":82,"updated_unix":28},"calf-integrated-arms-for-bimanual-quadruped-loco-manipulation","",{"@graph":36,"@context":77},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/calf-integrated-arms-for-bimanual-quadruped-loco-manipulation/84895/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-23","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"How is long-horizon skill sequencing achieved for autonomy?","Question",{"text":75,"@type":76},"A vision-language model sequences skills from a predefined library at each skill boundary, conditioned on the head-camera image and the task state, enabling long-horizon autonomy in simulation.","Answer","https://schema.org",{"og:url":51,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":84},[85,89,93,97,102,106,111,114,119,122,126],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":112,"slug":113},30,"research-report",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},9,"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":98,"slug":129},19,"General","general"]