{"paper":{"title":"Multistep Belief Space Dynamics Learning For Risk-Aware Control","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A structured multistep approach to learning distributional dynamics enables risk-aware MPC that naturally regulates speed in off-road driving.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Bogdan Vlahov, Evangelos A. Theodorou, Jason Gibson, Patrick Spieler","submitted_at":"2026-05-12T18:11:12Z","abstract_excerpt":"As autonomous vehicles move from a simplified research setting to practical use, there exists a large gap between the dynamic behavior of a human driving and an autonomous system. Risk-aware behavior needs to naturally develop in order to scale to the demands of the real world. A major issue for risk-aware planning and control has been predicting how dynamical uncertainty evolves through time and optimizing plans that account for this without being overly conservative. Here, we present a learning framework to predict distributional dynamics that can be optimized in real time for Model Predicti"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our planning architecture is able to naturally regulate the speed of the vehicle based on the environment and consistently demonstrates intelligent behavior over miles of diverse terrain.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That deviations from the proposed structure in learning distributional dynamics materially degrade MPC performance, and that the learned model generalizes beyond the specific off-road dataset without introducing hidden conservatism or instability.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A structured learning approach for multistep distributional dynamics in belief space enables real-time risk-aware MPC, validated via ablation on real off-road data and deployment on a full-sized vehicle.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A structured multistep approach to learning distributional dynamics enables risk-aware MPC that naturally regulates speed in off-road driving.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ba8df5783045ad00fe8dd4462f8a4e9c20d92171635071c0b7130b26ab405036"},"source":{"id":"2605.12628","kind":"arxiv","version":1},"verdict":{"id":"4807b1c3-d863-442f-9958-233d7edc1d6d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:41:08.745430Z","strongest_claim":"Our planning architecture is able to naturally regulate the speed of the vehicle based on the environment and consistently demonstrates intelligent behavior over miles of diverse terrain.","one_line_summary":"A structured learning approach for multistep distributional dynamics in belief space enables real-time risk-aware MPC, validated via ablation on real off-road data and deployment on a full-sized vehicle.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That deviations from the proposed structure in learning distributional dynamics materially degrade MPC performance, and that the learned model generalizes beyond the specific off-road dataset without introducing hidden conservatism or instability.","pith_extraction_headline":"A structured multistep approach to learning distributional dynamics enables risk-aware MPC that naturally regulates speed in off-road driving."},"references":{"count":58,"sample":[{"doi":"","year":null,"title":"A comprehensive review on autonomous navigation,","work_id":"d1cd195f-96ec-4d7b-a2e4-9b8e536487b1","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1145/3727642","year":null,"title":"Available: https://doi.org/10.1145/3727642 1","work_id":"3dd8fd14-71a6-49c3-aa7d-17d5c300fc15","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Parting with misconceptions about learning- based vehicle motion planning","work_id":"ca47adee-d361-4755-9f62-251c8a668963","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Quantifying generalization in reinforcement learning,","work_id":"76801677-183d-4fa1-a3b2-27ab614f9998","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"A survey on unmanned surface vehicles for disaster robotics: Main challenges and directions,","work_id":"f3384bcc-ada6-4b5b-bfa9-bebfdc5f980e","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":58,"snapshot_sha256":"2291ca454b9319c8a4ae9bee06d5bd3370bd68c62750980a98b4bf7d07a6734c","internal_anchors":1},"formal_canon":{"evidence_count":2,"snapshot_sha256":"0b57bfbc2b410eb5c6342cb5e5da5e64658a98ee2916b514d2d16a6c96782c76"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}