跳到主要导航 跳到搜索 跳到主要内容

Safety corridor and pedestrian prediction integrated sequential MPC for safe robot navigation in dynamic environments

  • Yinsong Qu
  • , Yunxiang Li
  • , Yudong Guo
  • , Yuezu Lv*
  • *此作品的通讯作者
  • Beijing Institute of Technology

科研成果: 期刊稿件 › 文章 › 同行评审

摘要

Safe and efficient obstacle avoidance is essential for mobile robot navigation in human-machine coexisting environments. This paper develops a sequential model predictive control (MPC) framework that integrates safety corridor generation and pedestrian trajectory prediction for stable and reliable autonomous navigation in complex dynamic scenarios. Static obstacles are constrained by multidirectional rectangular safety corridors optimized via Chebyshev circle screening, which improves free-space utilization and geometric adaptability to complex obstacle layouts. For dynamic pedestrians, a robust dynamic control barrier function module is embedded in the MPC optimizer to impose safe avoidance constraints. A neural-network-based detector extracts pedestrian bounding boxes from raw point cloud data, effectively alleviating target misdetection and track loss compared with traditional clustering methods. The established optimization model fuses static corridor constraints and dynamic pedestrian prediction to generate smooth, feasible, and safe robot motion commands. Gazebo simulations validate the promising navigation performance of the proposed method, which yields measurable improvements in corridor generation efficiency, trajectory smoothness, and travel efficiency under mixed static and dynamic obstacle environments.

源语言英语
期刊论文编号17298806261483759
期刊International Journal of Advanced Robotic Systems
卷23
期4
DOI
出版状态已出版 - 1 9月 2026
已对外发布是

学术指纹

探究 'Safety corridor and pedestrian prediction integrated sequential MPC for safe robot navigation in dynamic environments' 的科研主题。它们共同构成独一无二的学术指纹。

引用此