{"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/gated-path-planning-networks","title":"Gated Path Planning Networks","arxiv_id":"1806.06408","date":"2018-06-17","proceeding":"ICML 2018 7","authors":["Lisa Lee","Emilio Parisotto","Devendra Singh Chaplot","Eric Xing","Ruslan Salakhutdinov"],"abstract":"Value Iteration Networks (VINs) are effective differentiable path planning\nmodules that can be used by agents to perform navigation while still\nmaintaining end-to-end differentiability of the entire architecture. Despite\ntheir effectiveness, they suffer from several disadvantages including training\ninstability, random seed sensitivity, and other optimization problems. In this\nwork, we reframe VINs as recurrent-convolutional networks which demonstrates\nthat VINs couple recurrent convolutions with an unconventional max-pooling\nactivation. From this perspective, we argue that standard gated recurrent\nupdate equations could potentially alleviate the optimization issues plaguing\nVIN. The resulting architecture, which we call the Gated Path Planning Network,\nis shown to empirically outperform VIN on a variety of metrics such as learning\nspeed, hyperparameter sensitivity, iteration count, and even generalization.\nFurthermore, we show that this performance gap is consistent across different\nmaze transition types, maze sizes and even show success on a challenging 3D\nenvironment, where the planner is only provided with first-person RGB images.","url_abs":"http://arxiv.org/abs/1806.06408v1","url_pdf":"http://arxiv.org/pdf/1806.06408v1.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":"gated-path-planning-networks","repo_url":"https://github.com/lileee/gated-path-planning-networks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"gated-path-planning-networks","repo_url":"https://github.com/omron-sinicx/neural-astar","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"gated-path-planning-networks","repo_url":"https://github.com/omron-sinicx/planning-datasets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"sensitivity","task_name":"Sensitivity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.06408","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}