首页|基于嵌入式立体视觉的智能机器人障碍检测方法

基于嵌入式立体视觉的智能机器人障碍检测方法

扫码查看
针对嵌入式智能机器人移动过程中障碍物检测问题,提出了一种采用双激光扫描线辅助的双目立体视觉三维重建方法.该方法可以根据场景需要自适应地改变激光器的扫描角度和频率,动态地调整特征点的密度信息,使得DSP处理器在处理多任务时,能够有效解决特征匹配算法中关键场景深度信息不够的问题.在室内走廊行走测试实验中,该系统能成功检测到门的状态,并精确计算出门套及门槛的尺寸大小,该方法能有效提升嵌入式处理时效,距离误差小于5%.
Obstacle Detection Method of Intelligent Robot Based on Embedded Stereo Vision
Based on the obstacle detection problem of intelligent moving robot,this paper proposed a three-di-mensional reconstruction method of binocular stereo vision assisted by dual laser scanning lines.It can change the laser scanning angle and frequency self-adaptive according to the variety of the scene,and adjust the density in-formation of feature points dynamically.It can make the DSP processor to collect depth information efficiently at the feature matching algorithm with dealing multi tasks.According to the indoor corridor walking experiment,the system can detect the open door and calculate the parameters of the threshold.This method can improve the effec-tiveness of embedded processing,and the distance error is less than 5%.

intelligent robotstereo visionembedded ADSP-BF548feature matchstructured light

魏保龙、王霏

展开 >

宜春学院 数学与计算机科学学院,江西 宜春 336000

智能机器人 立体视觉 嵌入式ADSP-BF548 特征匹配 结构光

江西省教育厅科技项目

GJJ219309

2024

宜春学院学报
宜春学院

宜春学院学报

影响因子:0.271
ISSN:1671-380X
年,卷(期):2024.46(3)
  • 18