This website requires JavaScript.

A real-time dynamic obstacle tracking and mapping system for UAV navigation and collision avoidance with an RGB-D camera

Zhefan XuXiaoyang ZhanBaihan ChenYumeng XiuChenhao YangKenji Shimada
Sep 2022
摘要
The real-time dynamic environment perception has become vital for autonomousrobots in crowded spaces. Although the popular voxel-based mapping methods canefficiently represent 3D obstacles with arbitrarily complex shapes, they canhardly distinguish between static and dynamic obstacles, leading to the limitedperformance of obstacle avoidance. While plenty of sophisticated learning-baseddynamic obstacle detection algorithms exist in autonomous driving, thequadcopter's limited computation resources cannot achieve real-time performanceusing those approaches. To address these issues, we propose a real-time dynamicobstacle tracking and mapping system for quadcopter obstacle avoidance using anRGB-D camera. The proposed system first utilizes a depth image with anoccupancy voxel map to generate potential dynamic obstacle regions asproposals. With the obstacle region proposals, the Kalman filter and ourcontinuity filter are applied to track each dynamic obstacle. Finally, theenvironment-aware trajectory prediction method is proposed based on the Markovchain using the states of tracked dynamic obstacles. We implemented theproposed system with our custom quadcopter and navigation planner. Thesimulation and physical experiments show that our methods can successfullytrack and represent obstacles in dynamic environments in real-time and safelyavoid obstacles.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答