arXiv cs.AI· Rwik Rana, Jesse Quattrociocchi, Christian Ellis, Nathan Tsoi, Garrett Warnell, Joydeep Biswas·· 12 小时前AI 评分31
OptCar：将通用车辆动力学模型适配为跨地形高速 MPC 控制器
Adapting Generalist Vehicle Models for High-Speed MPC Across Terrains
AI 导读
OptCar 提出一套将通用前向动力学（FKD）基础模型特化为目标车辆专用模型的方法，采用 Transformer 架构并以 FiLM 将多步预测条件化于单个动力学上下文 token。
正文 · 原文
Abstract:High-speed off-road autonomy requires precise closed-loop control for a target vehicle while remaining robust across changing terrains. Recent forward kinodynamic (FKD) prediction foundation models suggest a promising path, starting from a generalist model and specializing it to the target platform. However, effective specialization remains challenging, as it often requires substantial real-world data, and models adapted to one setting can still overfit to specific terrains or driving regimes. We present OptCar (Optimized Car), a recipe for bridging the gap from generalist to specialist FKD models that preserves cross-terrain generalization while optimizing performance for a specific vehicle. OptCar introduces a transformer FKD architecture that uses FiLM to condition multi-step predictions on a single dynamics context token summarizing recent state-action history. It then specializes the generalist model using limited real-world data and targeted synthetic rollouts from environment-specific system identification. In closed-loop model predictive control (MPC) experiments across three terrains and an out-of-distribution cart-pulling task, the largest gains appear at 6 m/s, the highest speed evaluated and the regime in which slip dominates tracking error. On vegetation + dirt, the most slip-diverse terrain, OptCar reduces 6 m/s trajectory tracking error by roughly 55% relative to AnyCar fine-tuned on real data alone, and remains the most accurate even when an unseen cart payload changes the dynamics. With 5 minutes of real data per terrain, OptCar is competitive on road with a specialist trained on 30 minutes of road data and outperforms it when the terrain changes. Project Page : this https URL
| Comments: | this https URL Selected at CORL 2026 |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2607.13319 [cs.RO] |
| (or arXiv:2607.13319v4 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2607.13319 arXiv-issued DOI via DataCite |
Submission history
From: Rwik Rana [view email]
[v1]
Tue, 14 Jul 2026 22:59:40 UTC (21,516 KB)
[v2]
Wed, 7 Oct 2026 09:14:28 UTC (21,513 KB)
[v3]
Wed, 14 Oct 2026 06:06:18 UTC (21,537 KB)
[v4]
Fri, 16 Oct 2026 06:59:21 UTC (21,538 KB)
来源：arXiv cs.AI · arxiv.org