基于视觉的智能驾驶虚拟仿真平台与课程教学实践

Visual Based Virtual Simulation Platform of Intelligent Driving and Teaching Practice

基于视觉的智能驾驶作为自动化技术的重要应用成为当前研究的热点,但受到实验场景客观限制很难引入课堂的实践教学。为引导学生从理论出发、由实践深入,切实培养敢创新、善实践的复合型人才,构建了基于计算机视觉的智能驾驶虚拟仿真平台,在此基础上提出了一种感知环境深度信息的仿真算法,为解决实际场景标注大量样本的问题提供了解决方案;同时平台提供了多样化的智能驾驶仿真场景、多角度的观察视野以及丰富的功能控制模块。教学实践表明,该平台为学生系统地掌握计算机视觉领域知识,激发学习热情和培养创新实践能力起到了促进作用,平台也为实践教学提供了新的模式。

As an important application of automation technology,vision based intelligent driving becomes a research hotspot,but it is difficult to practice in classroom teaching due to the limitations of experimental scenarios.In order to guide students to start from theory,deepen from practice,and effectively cultivate compound talents who dare to innovate and practice well,this paper builds a virtual simulation platform for intelligent driving based on computer vision.On this basis,a simulation algorithm for perceiving the depth information of the environment is proposed,which provides a solution to the problem of labeling a large number of samples in actual scenes.At the same time,the platform provides a variety of intelligent driving simulation scenarios,multi-angle observation fields and rich functional control modules.Teaching practice shows that this platform plays a role in promoting students to systematically master the knowledge in the field of computer vision,stimulates students' enthusiasm for learning and cultivate innovative practical ability,and the platform also provides a new model for practical teaching.