[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84620-en":3,"doc-seo-84620-105":29,"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":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},84620,1649267921044,"Ava Thompson","https://us-avatar.wpscdn.com/avatar/1800007509477c92dfb?_k=1782875107921204101",8,"Research & Report","Simulation Based Reward Function Validation for Multi Agent On Orbit Inspection","Simulation based reward function validation for multi-agent orbital inspection uses multi-agent reinforcement learning to control groups of inspection spacecraft. Instead of restricting rewards to a fixed set of inspection points, the method generalizes the reward function using 3D reconstruction analysis of objects inspected in orbit. The framework evaluates arbitrary numbers of images at arbitrary locations and lets trained agents choose when to collect imagery. Reconstruction outcomes guide refinement for safe, fuel efficient collaboration and enable adjustable tradeoffs between reconstruction quality and data gathering effort.","arXiv :2607 .0 1367v 1 [ cs .MA] 1 Jul 2026  \nSimulation Based Reward Function Validation for Multi-Agent  \nOn Orbit Inspection  \nPatrick Quinn∗ , Bala Prenith Reddy Gopu†, George M. Nehma‡, and Dr. Madhur Tiwari §  \nFlorida Institute of Technology, Melbourne, FL., 32901  \nA proposed method for the control of groups of inspection spacecraft is Multi-Agent Reinforcement Learning (MARL). While MARL has already been employed for this purpose in previous work, the reward functions used focus on reaching a finite set of predetermined inspection points around the target. In this work, we study and develop a generalized reward function for the MARL inspection task informed by the analysis of 3D reconstructions of inspected objects in orbit. Because the reward function is generalized such that any number of images at arbitrary locations may evaluated, we also allow trained agents to have complete control over when images are collected. With this approach, we gather insights into best practices for not only the specific MARL inspection task, but also gain key takeaways informative to the broader inspection task outside of a MARL context.  \nI. Introduction  \nConcerns over the accumulation of space debris and the accompanying implications for the use of space in the future have led to calls for action to be taken to mitigate further debris build-up, as well as the active removal of debris already in orbit [1] . Prior to the removal of orbital debris, it is useful for missions to gather data on debris of interest, in order to properly allocate resources for their removal [2] . On these missions, imagery is often taken which can be used to generate 3D reconstructions of the target debris [3, 4] . The success of these missions is supportive of the usefulness and practicality of orbital inspection already, but some improvements may be necessary to make similar missions feasible when aiming to gather information on the large amount of debris in orbit. One of the issues in scaling up the scope of such missions is the amount of time and effort that goes into planning maneuvers for proximity operations with other objects in orbit, which can take teams of human operators multiple days to do manually [5] . Because of this, any methods which can be used to automate the control of the inspection spacecraft greatly increases the feasibility of larger missions. Additionally, rather than using one inspection spacecraft, it may be preferred to use multiple, cutting down on the time taken to accomplish the inspection task as well as adding redundancy to the mission [6] . Without a suitable automated control method, adding more spacecraft would only exacerbate the issues with mission planning already seen today. One proposed method for the control of a group of inspection spacecraft is the utilization of Multi-Agent Reinforcement Learning (MARL) . While MARL has already been employed for the spacecraft inspection task [7, 8], and other work has been done using Reinforcement Learning (RL) for the inspection of small bodies such as asteroids with a single spacecraft [9], the approaches taken so far have focused on learning policies to reach predetermined inspection points. The distribution of such inspection points, as well as the number of inspection points, is left as a design decision. A potential way to improve upon this approach is by defining reward functions which accommodate an arbitrary number of inspection points at arbitrary positions. In order to ensure the effectiveness and enable the refinement of such reward functions, they can be validated by using the trajectories generated by the MARL agents to gather images of the target in a high-fidelity simulation. Those images may be used to create 3D reconstructions of the target [10], which may then be compared to a reference model for the target to get both an objective and subjective sense for reward function performance. Shaping of the reward function can then be used to encourage behaviors in th","cbCaisXVPYyxCuIL","https://ap.wps.com/l/cbCaisXVPYyxCuIL","pdf",1992410,1,13,"English","en",105,"# I. Introduction\n# II. Theory","[{\"question\":\"How is reward function performance validated and refined?\",\"answer\":\"High-fidelity simulations generate agent trajectories and collect images, which are used to create 3D reconstructions. The reconstructions are compared to a reference model to assess reward performance, then used to refine the reward function for desired safety, fuel usage, collaboration, and quality–effort tradeoffs.\"}]",1784197187,33,{"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":78,"head_meta":80,"extra_data":82,"updated_unix":27},"simulation-based-reward-function-validation-for-multi-agent-on-orbit-inspection","",{"@graph":35,"@context":77},[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/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/simulation-based-reward-function-validation-for-multi-agent-on-orbit-inspection/84620/",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],{"name":72,"@type":73,"acceptedAnswer":74},"How is reward function performance validated and refined?","Question",{"text":75,"@type":76},"High-fidelity simulations generate agent trajectories and collect images, which are used to create 3D reconstructions. The reconstructions are compared to a reference model to assess reward performance, then used to refine the reward function for desired safety, fuel usage, collaboration, and quality–effort tradeoffs.","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":24},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":45,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":45,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":45,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":45,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]