{"paper":{"title":"Efficient Emotion-Aware Iconic Gesture Prediction for Robot Co-Speech","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A lightweight transformer predicts iconic gestures for robots from text and emotion alone, outperforming GPT-4o on the BEAT2 dataset.","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Christian Arzate Cruz, Edwin C. Montiel-Vazquez, Giorgos Giannakakis, Randy Gomez, Stefanos Gkikas, Thomas Kassiotis","submitted_at":"2026-04-13T13:02:02Z","abstract_excerpt":"Co-speech gestures increase engagement and improve speech understanding. Most data-driven robot systems generate rhythmic beat-like motion, yet few integrate semantic emphasis. To address this, we propose a lightweight transformer that derives iconic gesture placement and intensity from text and emotion alone, requiring no audio input at inference time. The model outperforms GPT-4o in both semantic gesture placement classification and intensity regression on the BEAT2 dataset, while remaining computationally compact and suitable for real-time deployment on embodied agents."},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The lightweight transformer outperforms GPT-4o in both semantic gesture placement classification and intensity regression on the BEAT2 dataset, while remaining computationally compact and suitable for real-time deployment on embodied agents.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That text and emotion labels alone are sufficient to accurately predict iconic gestures without audio cues, and that the BEAT2 dataset captures representative real-world co-speech behavior for this task.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Lightweight transformer predicts iconic gesture placement and intensity from text and emotion for robot co-speech, outperforming GPT-4o on BEAT2 without audio input.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A lightweight transformer predicts iconic gestures for robots from text and emotion alone, outperforming GPT-4o on the BEAT2 dataset.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e13555ac88faff9d6a167b6bc9b727932bb317d1b4b28eb79bc2d15f1b28fd12"},"source":{"id":"2604.11417","kind":"arxiv","version":4},"verdict":{"id":"400ba12e-c16c-4b60-93fc-490d5230b479","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T06:48:52.143415Z","strongest_claim":"The lightweight transformer outperforms GPT-4o in both semantic gesture placement classification and intensity regression on the BEAT2 dataset, while remaining computationally compact and suitable for real-time deployment on embodied agents.","one_line_summary":"Lightweight transformer predicts iconic gesture placement and intensity from text and emotion for robot co-speech, outperforming GPT-4o on BEAT2 without audio input.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That text and emotion labels alone are sufficient to accurately predict iconic gestures without audio cues, and that the BEAT2 dataset captures representative real-world co-speech behavior for this task.","pith_extraction_headline":"A lightweight transformer predicts iconic gestures for robots from text and emotion alone, outperforming GPT-4o on the BEAT2 dataset."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.11417/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":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}