Название: All Weather Robot Vision
Автор: Jiandong Tian
Издательство: Huazhong University of Science and Technology Press, Springer
Год: 2022
Страниц: 315
Язык: английский
Формат: pdf (true)
Размер: 14.7 MB
Research has shown that more than 70% of information that humans receive is perceived by vision. Similarly, vision systems are also very important for robots and have basically become the standard configuration of robots. It is fundamental for robots to understand the environment and execute tasks, and it is also an important entrance to realize robot intelligence. With the rapid development of Deep Learning (DL) technology in recent years, task-oriented researches in vision, such as visual servoing, target recognition, tracking, and image understanding, have made remarkable progress. However, it should be noted that these visual algorithms are all developed for images with good quality and stable lighting conditions. In fact, robots often need to work in complex, dynamic, or unstructured environment. At this time, vision algorithms often lack adaptive ability and robustness to the environment.