DBSCAN and Yolov5 based 3D object detection and its adaptation to a mobile platform

IF 3.1 3区计算机科学 Q2 AUTOMATION & CONTROL SYSTEMS

Mechatronics Pub Date : 2024-08-15 DOI:10.1016/j.mechatronics.2024.103238

引用次数: 0

Abstract

This study presents a 3D object detection technology for mobile platforms and its application. Rather than an innovative high-performance model, we proposed a “useable” model for the robot industry at the current technology stage by combining various techniques. To reduce computation time, a 2D region proposal was obtained using a RGB image-based CNN model. By applying the DBSCAN clustering technique to the point cloud corresponding to the 2D region proposal, a method of obtaining a 3D region proposal was proposed. This allowed for 3D object detection using an RGB image dataset, which has been widely researched, while reducing the computation load to a level suitable for use in mobile robots. Furthermore, the 3D object detection was integrated into a ROS 2-based mobile platform, which was used to perform pedestrian-safe avoidance tasks and elevator button operation tasks. The performance was confirmed through experiments.

查看原文本刊更多论文

基于 DBSCAN 和 Yolov5 的 3D 物体检测及其与移动平台的适配

本研究介绍了用于移动平台的三维物体检测技术及其应用。与其说我们提出了一个创新的高性能模型，不如说我们结合了各种技术，为机器人行业提出了一个在当前技术阶段 "可用 "的模型。为了减少计算时间，我们使用基于 RGB 图像的 CNN 模型获得了二维区域建议。通过将 DBSCAN 聚类技术应用于与二维区域提案相对应的点云，提出了一种获得三维区域提案的方法。这样，就可以使用已被广泛研究的 RGB 图像数据集进行三维物体检测，同时将计算负荷降低到适合移动机器人使用的水平。此外，三维物体检测被集成到基于 ROS 2 的移动平台中，用于执行行人安全避让任务和电梯按钮操作任务。实验证实了其性能。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Mechatronics 工程技术-工程：电子与电气

CiteScore

5.90

自引率

9.10%

发文量

审稿时长

109 days

期刊介绍： Mechatronics is the synergistic combination of precision mechanical engineering, electronic control and systems thinking in the design of products and manufacturing processes. It relates to the design of systems, devices and products aimed at achieving an optimal balance between basic mechanical structure and its overall control. The purpose of this journal is to provide rapid publication of topical papers featuring practical developments in mechatronics. It will cover a wide range of application areas including consumer product design, instrumentation, manufacturing methods, computer integration and process and device control, and will attract a readership from across the industrial and academic research spectrum. Particular importance will be attached to aspects of innovation in mechatronics design philosophy which illustrate the benefits obtainable by an a priori integration of functionality with embedded microprocessor control. A major item will be the design of machines, devices and systems possessing a degree of computer based intelligence. The journal seeks to publish research progress in this field with an emphasis on the applied rather than the theoretical. It will also serve the dual role of bringing greater recognition to this important area of engineering.