arXiv：cs.LG· Seunghwan Jang, Jeongyong Yang, Siddharth Ancha, SooJean Han·· 3 小时前AI 评分33
SafeStreamingFlow：让采样动态对齐执行动态的安全流规划
Safe Streaming Flow Planning by Aligning Sampling Dynamics with Execution Dynamics
AI 导读
SafeStreamingFlow 是一种目标条件规划器，通过将流采样动态与执行动态对齐，并用高阶控制屏障函数仅对已执行步施加安全约束，从而降低规划延迟并提升安全性。该方法在导航、竞速和运动基准上相比现有安全扩散/流规划器减少了规划延迟、改善了安全性，同时保持有竞争力的目标到达成功率。该工作已被 CoRL 2026 接收。
正文
Abstract:Generative planners based on diffusion/flow matching can learn to synthesize long-horizon trajectories from demonstrations. However, real-world deployment requires (i) enforcing safety constraints during execution and (ii) tight online replanning at fast execution rates. Prior safe diffusion/flow planners generate the agent's full trajectory at once, while repeatedly perturbing intermediate states to satisfy safety constraints. This approach is not only computationally intensive, but also introduces distribution shift since the learned sampling dynamics is distinct from the system's execution dynamics. We propose SafeStreamingFlow, a goal-conditioned planner that aligns flow sampling dynamics with execution dynamics by sequentially integrating a learned state vector field with hierarchical state prediction. Importantly, we need to enforce safety constraints only for the executed step via high order control barrier functions. Across navigation, racing, and locomotion benchmarks, SafeStreamingFlow reduces planning latency and improves safety compared to existing methods, while maintaining competitive goal-reaching success.
| Comments: | Accepted to the 10th Conference on Robot Learning (CoRL 2026). Project page: this https URL |
| Subjects: | Robotics (cs.RO); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.03132 [cs.RO] |
| (or arXiv:2610.03132v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03132 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Seunghwan Jang [view email]
[v1]
Fri, 2 Oct 2026 10:52:47 UTC (2,382 KB)
来源：arXiv：cs.LG · arxiv.org