{"paper":{"title":"MUJICA: Multi-skill Unified Joint Integration of Control Architecture for Wheeled-Legged Robots","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A single policy integrates omnidirectional moving, climbing, and fall recovery for wheeled-legged robots using only proprioceptive sensing.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Lihua Zhang, Peng Zhai, Qianxiang Yu, Quancheng Qian, Xiaoyi Wei, Yueqi Zhang, Yuqi Li, Zhengxu He","submitted_at":"2026-05-13T06:31:18Z","abstract_excerpt":"Wheeled-legged robots hold promise for traversing complex terrains and offer superior mobility compared to legged robots. However, wheeled-legged robots must effectively balance both wheeled driving and legged control. Furthermore, due to noisy proprioceptive sensing and real-world motor constraints, realizing robust and adaptive locomotion at peak performance of motors remains challenging. We propose the Multi-skill Unified Joint Integration of Control Architecture (MUJICA), a unified, fully proprioceptive control framework for wheeled-legged robots that integrates diverse low-level skills-in"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We propose the Multi-skill Unified Joint Integration of Control Architecture (MUJICA), a unified, fully proprioceptive control framework for wheeled-legged robots that integrates diverse low-level skills—including omnidirectional moving, high platform climbing, and fall recovery—within a single policy.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That joint training with indicator variables and accurate DC-motor modeling in simulation is sufficient to produce robust sim-to-real transfer and reliable proprioceptive skill selection without additional real-world adaptation or external sensing.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A single reinforcement learning policy jointly trains multiple locomotion skills for wheeled-legged robots with DC-motor constraints and learns a proprioceptive skill selector for adaptive behavior.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A single policy integrates omnidirectional moving, climbing, and fall recovery for wheeled-legged robots using only proprioceptive sensing.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"926c4f0ba1fbc1a5539512ffb1a990f5525360e5154065e5cc2b608583221020"},"source":{"id":"2605.13058","kind":"arxiv","version":1},"verdict":{"id":"1ff8071b-0694-4aac-bf24-ae3f521339fd","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:08:31.581622Z","strongest_claim":"We propose the Multi-skill Unified Joint Integration of Control Architecture (MUJICA), a unified, fully proprioceptive control framework for wheeled-legged robots that integrates diverse low-level skills—including omnidirectional moving, high platform climbing, and fall recovery—within a single policy.","one_line_summary":"A single reinforcement learning policy jointly trains multiple locomotion skills for wheeled-legged robots with DC-motor constraints and learns a proprioceptive skill selector for adaptive behavior.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That joint training with indicator variables and accurate DC-motor modeling in simulation is sufficient to produce robust sim-to-real transfer and reliable proprioceptive skill selection without additional real-world adaptation or external sensing.","pith_extraction_headline":"A single policy integrates omnidirectional moving, climbing, and fall recovery for wheeled-legged robots using only proprioceptive sensing."},"references":{"count":31,"sample":[{"doi":"","year":2025,"title":"Deep reinforcement learning for robotics: A survey of real- world successes,","work_id":"3f56a944-0070-4884-83f8-ce79b55e7153","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Moe-loco: Mixture of experts for multitask locomotion","work_id":"74ccf560-fcd2-4440-8d49-7dfcd7c78aec","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Deep reinforcement learning in mixture of experts control system for blind wheeled-legged quadrupedal locomo- tion,","work_id":"db9466cc-6ea7-4c4c-a0b0-9d4fc2139274","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Dreamwaq: Learning robust quadrupedal locomotion with implicit terrain imagination via deep reinforcement learning","work_id":"0621d93a-2a38-40a5-ac3c-f1e892bafc18","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Fr-net: Learning robust quadrupedal fall recovery on challenging terrains through mass-contact prediction,","work_id":"4e68d29c-ea21-4127-b6ee-d126532f9cd5","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":31,"snapshot_sha256":"e84e8b6295e7197cca045bdf5a9fc67c52da40d69365a1b0085dc9563f7b3c2b","internal_anchors":5},"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"}