跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.05793 · v2

ZipDrift: Generating Multimodal Short-Horizon MPC Costs with One-Step Drifting Models

Ke Zhang
cs.ROcs.LGeess.SY
首次提交 2026-11-04 08:58
版本更新 2026-11-05 12:44
雷达观测 2026-11-06 11:06
arXiv ↗PDF ↗

摘要

Short-horizon model predictive control (MPC) is fast but myopic. ZipMPC compensates by learning a context-dependent correction of the short-horizon cost that imitates a long-horizon expert. We observe that the optimal correction is often multimodal: when a slower opponent can be passed on either side, good corrections of opposite sign coexist, and the expert's choice between them hinges on tiny cost differences. Deterministic regression then averages modes or inherits inconsistent expert tie-breaking. We propose ZipDrift, which trains a one-step drifting generator over MPC cost corrections, tilted toward low closed-loop cost and learned only from cost evaluations of candidate corrections, without differentiating through the optimiser. A few generated candidates are ranked by a learned critic, and the model-based MPC then enforces dynamics and constraints. In a kinematic overtaking benchmark derived from ZipMPC, more than half of the contexts are bimodal and the manual short-horizon MPC crashes in 28% of them. With an eight-dimensional stagewise correction, evaluated on 1000 fresh contexts and five training seeds, ZipDrift with eight candidates crashes in 0.74% of episodes, compared with 3.4% for an idealised ZipMPC-style regressor and 2.1% for critic-ranked random candidates with the same budget. A mixture density network (MDN) with the same selection is equally good (0.76%). The benefit of drifting lies in sample quality: in a pre-specified test on further unseen contexts, a single drifting sample crashes in 8.7% of episodes against 13.2% for the MDN, and ZipDrift needs four candidates instead of eight to fall below 1%. With many candidates, all methods degrade as selection exploits errors of the critic.

作者备注：7 pages, 3 figures, 5 tables. Preliminary simulation study

查看 arXiv 原文 ↗阅读 PDF ↗

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存，登录后共享已有结果。

版本差异

仅对比实际观测的元数据，不代表 PDF 正文或实验变化。版本可能不连续；带删除线为移除，下划线为新增。

摘要变化

Short-horizon model predictive control (MPC) is fast but myopic. ZipMPC compensates by learning a context-dependent correction of the short-horizon cost that imitates a long-horizon expert. We observe that the optimal correction is often multimodal: when a slower opponent can be passed on either side, good corrections of opposite sign coexist, and the expert's choice between them hinges on tiny cost differences. Deterministic regression then averages modes or inherits inconsistent expert tie-breaking. We propose ZipDrift, which trains a one-step drifting generator over MPC cost corrections, tilted toward low closed-loop cost and learned only from cost evaluations of candidate corrections, without differentiating through the optimiser. A few generated candidates are ranked by a learned critic, and the model-based MPC then enforces dynamics and constraints. In a kinematic overtaking benchmark derived from ZipMPC, 53%more than half of the contexts are bimodal and the manual short-horizon MPC crashes in 28.5% of test episodesthem. With an eight-dimensional stagewise correction, ZipDrift is collision-free inevaluated allon 6001000 testfresh episodescontexts and matches the bestfive evaluatedtraining candidateseeds, whereas anZipDrift idealisedwith ZipMPC-styleeight regressorcandidates crashes in 4.0.74% (16of episodes, compared with 3.04% withfor inconsistentan expertidealised labels)ZipMPC-style regressor and 2.1% for critic-ranked random candidates with the same budget. A mixture density network (MDN) with the same selection is equally good (0.76%). The benefit of drifting lies in 2sample quality: in a pre-specified test on further unseen contexts, a single drifting sample crashes in 8.07% of episodes against 13.2% for the MDN, and ZipDrift needs four candidates instead of eight to fall below 1%. With many candidates, all methods degrade as selection exploits errors of the critic.

作者备注变化

67 pages, 23 figures, 45 tables. Preliminary simulation study

观测到的版本

相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-05 12:44 · 最新观测版本

ZipDrift: Generating Multimodal Short-Horizon MPC Costs with One-Step Drifting Models

Short-horizon model predictive control (MPC) is fast but myopic. ZipMPC compensates by learning a context-dependent correction of the short-horizon cost that imitates a long-horizon expert. We observe that the optimal correction is often multimodal: when a slower opponent can be passed on either side, good corrections of opposite sign coexist, and the expert's choice between them hinges on tiny cost differences. Deterministic regression then averages modes or inherits inconsistent expert tie-breaking. We propose ZipDrift, which trains a one-step drifting generator over MPC cost corrections, tilted toward low closed-loop cost and learned only from cost evaluations of candidate corrections, without differentiating through the optimiser. A few generated candidates are ranked by a learned critic, and the model-based MPC then enforces dynamics and constraints. In a kinematic overtaking benchmark derived from ZipMPC, more than half of the contexts are bimodal and the manual short-horizon MPC crashes in 28% of them. With an eight-dimensional stagewise correction, evaluated on 1000 fresh contexts and five training seeds, ZipDrift with eight candidates crashes in 0.74% of episodes, compared with 3.4% for an idealised ZipMPC-style regressor and 2.1% for critic-ranked random candidates with the same budget. A mixture density network (MDN) with the same selection is equally good (0.76%). The benefit of drifting lies in sample quality: in a pre-specified test on further unseen contexts, a single drifting sample crashes in 8.7% of episodes against 13.2% for the MDN, and ZipDrift needs four candidates instead of eight to fall below 1%. With many candidates, all methods degrade as selection exploits errors of the critic.

查看此版本解读 →查看此版本原文 ↗
v1 · 2026-11-04 08:58

ZipDrift: Generating Multimodal Short-Horizon MPC Costs with One-Step Drifting Models

Short-horizon model predictive control (MPC) is fast but myopic. ZipMPC compensates by learning a context-dependent correction of the short-horizon cost that imitates a long-horizon expert. We observe that the optimal correction is often multimodal: when a slower opponent can be passed on either side, good corrections of opposite sign coexist, and the expert's choice between them hinges on tiny cost differences. Deterministic regression then averages modes or inherits inconsistent expert tie-breaking. We propose ZipDrift, which trains a one-step drifting generator over MPC cost corrections, tilted toward low closed-loop cost and learned only from cost evaluations of candidate corrections, without differentiating through the optimiser. A few generated candidates are ranked by a learned critic, and the model-based MPC then enforces dynamics and constraints. In a kinematic overtaking benchmark derived from ZipMPC, 53% of contexts are bimodal and the manual short-horizon MPC crashes in 28.5% of test episodes. With an eight-dimensional stagewise correction, ZipDrift is collision-free in all 600 test episodes and matches the best evaluated candidate, whereas an idealised ZipMPC-style regressor crashes in 4.0% (16.0% with inconsistent expert labels) and critic-ranked random candidates with the same budget in 2.0%.

查看此版本解读 →查看此版本原文 ↗