{"paper":{"title":"Goal-Conditioned Decision Transformer for Multi-Goal Offline Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A goal-conditioned Decision Transformer learns multi-goal robotics policies from offline data alone.","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Dominik \\.Zurek, Kamil Faber, Marcin Pietro\\'n, Pawe{\\l} Gajewski","submitted_at":"2024-10-08T20:35:30Z","abstract_excerpt":"Reinforcement learning (RL) in robotics faces significant hurdles regarding sample efficiency and generalization across varying goals. While Offline RL mitigates the need for costly online interactions, its integration with goal-conditioned policies and transformer-based architectures remains underexplored. We introduce a Goal-Conditioned Decision Transformer adapted for offline multi-goal robotics. By explicitly incorporating goal states into the sequence modeling framework, our approach efficiently solves varying tasks using only pre-collected data. We validate this method on a newly release"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experimental results demonstrate that our approach outperforms state-of-the-art online baselines in complex tasks and maintains robustness in sparse-reward settings, even with limited expert demonstrations.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The newly released offline dataset for the Franka Emika Panda platform contains sufficient coverage of varying goals and task distributions to support generalization of the goal-conditioned policy.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A Goal-Conditioned Decision Transformer is adapted for offline multi-goal RL and shown to outperform online baselines on a new Franka Emika Panda dataset.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A goal-conditioned Decision Transformer learns multi-goal robotics policies from offline data alone.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"46aee2f2464f78695ec23e481e1fa5db33cea3b711f125768d5b6025d0fd5653"},"source":{"id":"2410.06347","kind":"arxiv","version":2},"verdict":{"id":"763aed6a-5d64-4f01-af8a-e6561eaee0ca","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-23T19:17:55.724494Z","strongest_claim":"Experimental results demonstrate that our approach outperforms state-of-the-art online baselines in complex tasks and maintains robustness in sparse-reward settings, even with limited expert demonstrations.","one_line_summary":"A Goal-Conditioned Decision Transformer is adapted for offline multi-goal RL and shown to outperform online baselines on a new Franka Emika Panda dataset.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The newly released offline dataset for the Franka Emika Panda platform contains sufficient coverage of varying goals and task distributions to support generalization of the goal-conditioned policy.","pith_extraction_headline":"A goal-conditioned Decision Transformer learns multi-goal robotics policies from offline data alone."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2410.06347/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":47,"sample":[{"doi":"","year":2023,"title":"Continuous improvement of self-driving cars using dynamic confidence-aware reinforcement learning","work_id":"3118549e-dd15-4a2b-b12a-24dea48f5f49","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Efficient reinforcement learning for autonomous driving with parameterized skills and priors","work_id":"1d307660-5b4d-4a1b-9a8b-d2cafc8ade99","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Accelerating reinforcement learning for autonomous driving using task-agnostic and ego-centric motion skills","work_id":"f96e0b91-27d5-4a91-87e5-7a028e7b697a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"A review paper on implementing reinforcement learning technique in optimising games performance","work_id":"1d3d4815-5b61-43f0-b520-a2861ad6e203","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Modeling decisions in games using reinforcement learning","work_id":"b17fd2f8-7b44-4acd-8f9d-0d3a31fbbc54","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":47,"snapshot_sha256":"7d94eb00448a5b2243b6055c3b3968c887f6570405f846d3497ceff8540aa7c5","internal_anchors":4},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}