{"paper":{"title":"CoLA-Flow Policy: Temporally Coherent Imitation Learning via Continuous Latent Action Flow Matching for Robotic Manipulation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Performing flow matching inside a continuous latent action space produces near-single-step inference and markedly smoother robotic trajectories than direct action-space methods.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Jiang Zhiduo, Liu Hong, Liu Yang, Sun Wandong, Wu Songwei, Xie Guanghu, Zhao Rui","submitted_at":"2026-01-30T15:36:43Z","abstract_excerpt":"Learning long-horizon robotic manipulation requires jointly achieving expressive behavior modeling, real-time inference, and stable execution, which remains challenging for existing generative policies. Diffusion-based approaches offer strong modeling capacity but incur high inference latency, while flow matching enables fast, near-single-step generation yet often suffers from unstable execution when operating directly in the raw action space. We propose Continuous Latent Action Flow Policy (CoLA-Flow Policy), a trajectory-level imitation learning framework that performs flow matching in a con"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"CoLA-Flow Policy achieves near-single-step inference, improves trajectory smoothness by up to 93.7% and task success by up to 25 percentage points over raw action-space flow baselines, while remaining significantly faster than diffusion-based policies.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That mapping action sequences into a continuous latent space and learning an explicit latent flow will reliably separate global motion structure from low-level control noise across varied real-world conditions and sensor modalities.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CoLA-Flow Policy encodes action sequences into latent trajectories and performs flow matching there, yielding near-single-step inference with up to 93.7% smoother trajectories and 25-point higher success rates than raw-action flow baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Performing flow matching inside a continuous latent action space produces near-single-step inference and markedly smoother robotic trajectories than direct action-space methods.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"61ff2559757420cfdbeb856b37388c4652e99abca0872b5ccd223a3d5770759e"},"source":{"id":"2601.23087","kind":"arxiv","version":4},"verdict":{"id":"37b9ea98-e446-4308-8941-ee8f4584e041","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T09:19:32.024297Z","strongest_claim":"CoLA-Flow Policy achieves near-single-step inference, improves trajectory smoothness by up to 93.7% and task success by up to 25 percentage points over raw action-space flow baselines, while remaining significantly faster than diffusion-based policies.","one_line_summary":"CoLA-Flow Policy encodes action sequences into latent trajectories and performs flow matching there, yielding near-single-step inference with up to 93.7% smoother trajectories and 25-point higher success rates than raw-action flow baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That mapping action sequences into a continuous latent space and learning an explicit latent flow will reliably separate global motion structure from low-level control noise across varied real-world conditions and sensor modalities.","pith_extraction_headline":"Performing flow matching inside a continuous latent action space produces near-single-step inference and markedly smoother robotic trajectories than direct action-space methods."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2601.23087/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":3,"snapshot_sha256":"a17a8d15a1840ee98955022b2ffd4612c9b6cbe178e2235ec06bcbf4aa80f334"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}