[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82806-en":3,"doc-seo-82806-105":29,"detail-sidebar-cat-0-en-105":90},{"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":11,"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},82806,4398048950312,"Violet","https://ap-avatar.wpscdn.com/avatar/400002538284de19e3c?_k=1778320343897328908",8,"Research & Report","A Perception Manipulation Robotics System for Food Cutting","A perception-manipulation robotics framework for food cutting focuses on selecting an appropriate knife and adapting cutting behavior to diverse food properties. The system begins with a fixed trial cut that collects force measurements to infer hardness and friction-relevant characteristics, then chooses between a fruit knife and a serrated knife. An on-policy reinforcement learning controller optimizes cutting speed versus energy efficiency during an adaptive cutting phase. Experiments show 100% success on unseen foods and performance comparable to human operators.","A Perception-Manipulation Robotics System for Food Cutting  \nXinyuan Luo 1 , Wenzhen Yuan 1  \narXiv :2607 .04367v 1 [ cs .RO] 5 Jul 2026  \nAbstract—In the development of cooking robots, mastering the task of cutting is crucial. A significant challenge lies in the diverse properties of food, which necessitate distinct cutting policies and even different knives for optimal processing. This paper presents a perception-manipulation framework for foodcutting tasks. Our system features a knife selection module that utilizes force data from a preliminary fixed trial cut to select the appropriate knife for the given food. This is followed by an adaptive cutting phase using reinforcement learning (RL) to balance cutting speed and energy efficiency. In our experiments, the knife selection module achieved 100% successful rate on unseen food, and we compared the performances of fixed policy, RL policy, with human operators. Our method not only achieves high performance but also demonstrates comparable results to those of human participants. Project page: [https:](https:)//[robotproject8.github.io/robocutting/](robotproject8.github.io/robocutting/)  \nI. INTRODUCTION  \nThe development of cooking robots has been a key interest among researchers and industry practitioners for many years. The comprehensive cooking process includes food preparation tasks such as cutting [1], peeling [2], and stirring [3], along with various cooking techniques like flipping [4] and stir-frying [5] . Most of these sub-tasks present significant challenges due to the diverse properties and irregular shapes of food, as well as the delicate manipulation skills required.  \nCutting, a critical process in food preparation, necessitates a thorough understanding of food properties such as hardness and friction coefficient. While humans have an innate sense of these properties before cutting, they also adapt their cutting techniques in real-time based on the force-torque feedback received from knife contact. This adaptation control can be achieved by learning a dynamic model for model predictive control (MPC) [1], [6], [7], employing preset control algorithms [8]–[10], and utilizing reinforcement learning (RL) [11], [12] .  \nIn this work, we aim to develop a food-cutting system that generalizes to various types of unseen foods. Similar to humans, our robot initially interacts with the food to assess its properties. During the cutting process, the robot can dynamically adjust its cutting strategy based on force feedback. Fig. 1 provides an overview of the proposed system. To expand the variety of foods it can handle, we have equipped the robot with an easy-to-switch knife mount that allows seamless transitions between a fruit knife and a serrated knife. As shown in Fig 2, the whole process begins with a fixed trial cut to collect force data, which is then used to determine the appropriate knife. Then, we have designed  \n1Xinyuan Luo and Wenzhen Yuan are with the University of Illinois at Urbana-Champaign {xl153, [yuanwz](yuanwz}@illinois.edu)[}](yuanwz}@illinois.edu)[@illinois.edu](yuanwz}@illinois.edu)  \nFig. 1: We developed a food cutting system that first performs a fixed trial cut motion to the food to understand the food properties and select a proper knife. Following this, the food is cut using an RL adaptive controller.  \na reward system to measure cutting efficiency and employed an on-policy RL agent to online refine the cutting motions.  \nA core challenge in food cutting stems from the diverse properties of food, such as hardness, surface friction coefficient, and juiciness. For instance, chopping is unsuitable for baguette due to its deformable nature; instead, a serrated knife and a sawing motion are required. Additionally, cutting juicy foods like tomatoes too hard may lead to damage due to their delicate structure [13], [14] . To tackle this challenge, we designed an easy-to-switch knife mount shown in Fig 4(c) that could effectively switch between a fruit knif","cbCais3VnLeW13pt","https://ap.wps.com/l/cbCais3VnLeW13pt","pdf",2897653,2,1,"English","en",105,"# Introduction\n## Food cutting challenges and properties\n## Proposed perception-manipulation pipeline\n## Adaptive reinforcement learning design\n# Related Works\n## Robotics cutting","[{\"question\":\"How does the system choose the right knife for a given food?\",\"answer\":\"It performs an initial fixed trial cut to collect force data, and then uses the measurements to select the appropriate knife for the target food.\"},{\"question\":\"What role does reinforcement learning play in the cutting process?\",\"answer\":\"After knife selection, an on-policy reinforcement learning agent adaptively refines cutting motions to balance cutting speed and energy efficiency.\"},{\"question\":\"How is the robot expected to handle different and unseen foods?\",\"answer\":\"The trial cut provides information about food properties, and the adaptive controller is trained to generalize across various types, achieving 100% success on four kinds of unseen food and comparable results to human operators.\"}]",1784183079,20,{"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":85,"head_meta":87,"extra_data":89,"updated_unix":27},"a-perception-manipulation-robotics-system-for-food-cutting","",{"@graph":35,"@context":84},[36,52,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,46,49],{"item":40,"name":41,"@type":42,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":20},"https://docshare.wps.com/document/","Document",{"item":47,"name":12,"@type":42,"position":48},"https://docshare.wps.com/document/research-report/",3,{"item":50,"name":13,"@type":42,"position":51},"https://docshare.wps.com/document/a-perception-manipulation-robotics-system-for-food-cutting/82806/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"inLanguage":23,"description":14,"dateModified":60,"datePublished":61,"encodingFormat":59,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":40,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-07-20","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"How does the system choose the right knife for a given food?","Question",{"text":74,"@type":75},"It performs an initial fixed trial cut to collect force data, and then uses the measurements to select the appropriate knife for the target food.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What role does reinforcement learning play in the cutting process?",{"text":79,"@type":75},"After knife selection, an on-policy reinforcement learning agent adaptively refines cutting motions to balance cutting speed and energy efficiency.",{"name":81,"@type":72,"acceptedAnswer":82},"How is the robot expected to handle different and unseen foods?",{"text":83,"@type":75},"The trial cut provides information about food properties, and the adaptive controller is trained to generalize across various types, achieving 100% success on four kinds of unseen food and comparable results to human operators.","https://schema.org",{"og:url":50,"og:type":86,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":88,"canonical":50},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,126,129,133],{"id":21,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":51,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]