{"paper":{"title":"Coordinating Multiple Conditions for Trajectory-Controlled Human Motion Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"CMC coordinates text and trajectory conditions via two-stage diffusion to generate accurate human motions","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Changxing Ding, Deli Cai, Haoyang Ma","submitted_at":"2026-05-13T16:09:04Z","abstract_excerpt":"Trajectory-controlled human motion generation aims to synthesize realistic human motions conditioned on both textual descriptions and spatial trajectories. However, existing methods suffer from two critical limitations: first, the conflict between text and trajectory conditions disrupts the denoising process, resulting in compromised motion quality or inaccurate trajectory following; second, the use of redundant motion representations introduces inconsistencies between motion components, leading to instability during trajectory control. To address these challenges, we propose CMC, a decoupled "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on HumanML3D and KIT datasets demonstrate that CMC achieves state-of-the-art performance in control accuracy and motion quality.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The simplified controlled-joint representation produced by the first-stage trajectory model supplies sufficient partial observations for the second-stage text-conditioned inpainting model to generate consistent, artifact-free full-body motions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CMC decouples trajectory control and text-conditioned motion completion with selective inpainting to achieve state-of-the-art accuracy and quality in multimodal human motion generation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"CMC coordinates text and trajectory conditions via two-stage diffusion to generate accurate human motions","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5b45bb0b3988ad8996403aea2d1ecbcf8cbfb461a27fb3b8869b211238b4c3bb"},"source":{"id":"2605.13729","kind":"arxiv","version":1},"verdict":{"id":"55ec01d5-2024-4269-839e-15162a4b6784","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:53:12.336054Z","strongest_claim":"Experiments on HumanML3D and KIT datasets demonstrate that CMC achieves state-of-the-art performance in control accuracy and motion quality.","one_line_summary":"CMC decouples trajectory control and text-conditioned motion completion with selective inpainting to achieve state-of-the-art accuracy and quality in multimodal human motion generation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The simplified controlled-joint representation produced by the first-stage trajectory model supplies sufficient partial observations for the second-stage text-conditioned inpainting model to generate consistent, artifact-free full-body motions.","pith_extraction_headline":"CMC coordinates text and trajectory conditions via two-stage diffusion to generate accurate human motions"},"references":{"count":60,"sample":[{"doi":"","year":2026,"title":"Crowdmogen: Event-driven collective human motion generation.Int","work_id":"f0de9a9c-017d-401a-a14d-47a147682da0","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Executing your commands via motion diffusion in latent space","work_id":"a5254f7b-dfc0-4712-a26d-ea5cb6eef6c8","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Hop: Heterogeneous topology-based multimodal entanglement for co-speech gesture generation","work_id":"8e563704-3e57-4e5a-b747-c40790f756ac","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Interaction transformer for human reaction generation.IEEE Trans","work_id":"ba5740ee-e57e-4166-ad49-60b5b7cc409e","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Mofusion: A framework for denoising-diffusion- based motion synthesis","work_id":"93509eac-e67d-4563-a1fc-ab1ee8180664","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":60,"snapshot_sha256":"c216d851a680a0b7ce27da01f0b87e9d0bba7e690117fa8daefd94a9eb3348e1","internal_anchors":0},"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"}