{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/bootstrapping-robotic-ecological-perception","title":"Bootstrapping Robotic Ecological Perception from a Limited Set of Hypotheses Through Interactive Perception","arxiv_id":"1901.10968","date":"2019-01-30","proceeding":null,"authors":["Léni K. Le Goff","Ghanim Mukhtar","Alexandre Coninx","Stéphane Doncieux"],"abstract":"To solve its task, a robot needs to have the ability to interpret its\nperceptions. In vision, this interpretation is particularly difficult and\nrelies on the understanding of the structure of the scene, at least to the\nextent of its task and sensorimotor abilities. A robot with the ability to\nbuild and adapt this interpretation process according to its own tasks and\ncapabilities would push away the limits of what robots can achieve in a non\ncontrolled environment. A solution is to provide the robot with processes to\nbuild such representations that are not specific to an environment or a\nsituation. A lot of works focus on objects segmentation, recognition and\nmanipulation. Defining an object solely on the basis of its visual appearance\nis challenging given the wide range of possible objects and environments.\nTherefore, current works make simplifying assumptions about the structure of a\nscene. Such assumptions reduce the adaptivity of the object extraction process\nto the environments in which the assumption holds. To limit such assumptions,\nwe introduce an exploration method aimed at identifying moveable elements in a\nscene without considering the concept of object. By using the interactive\nperception framework, we aim at bootstrapping the acquisition process of a\nrepresentation of the environment with a minimum of context specific\nassumptions. The robotic system builds a perceptual map called relevance map\nwhich indicates the moveable parts of the current scene. A classifier is\ntrained online to predict the category of each region (moveable or\nnon-moveable). It is also used to select a region with which to interact, with\nthe goal of minimizing the uncertainty of the classification. A specific\nclassifier is introduced to fit these needs: the collaborative mixture models\nclassifier. The method is tested on a set of scenarios of increasing\ncomplexity, using both simulations and a PR2 robot.","url_abs":"http://arxiv.org/abs/1901.10968v1","url_pdf":"http://arxiv.org/pdf/1901.10968v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"bootstrapping-robotic-ecological-perception","repo_url":"https://github.com/LeniLeGoff/IAGMM_Lib","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1901.10968","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}