Visual SLAM for automated industrial robotic cleaners

点击翻译按钮获取中文摘要

Cyberworks Robotics is an Ontario company that has been making robotic systems for over 30 years. Primarily, Cyberworks sold robot guidance systems that are mounted on the industrial cleaning machines. Traditionally, Cyberworks used ultrasonic sensing as the primary means of seeing the environment. These types of sensors suffer certain limitations, such as not being able to distinguish similar shaped objects from one another. The recent introduction of commodity rgb-d sensors (e.g., kinect camera) provides an opportunity significantly improve their robotic guidance systems. In order for a robotic cleaner to clean, it needs Map out its environment and keep track of where it is relative to its environment - called SLAM (Simultaneous Localization and Mapping) Keep track of where it has been and minimize revisiting the same floor space, also called Path Planning Clean within 3 cm of all objects For this project, we will investigate these automation problems within an industrial cleaning setting, such as office buildings, warehouses and hospitals. With the better, in depth understanding , we will then develop combined visual SLAM and coverage algorithm. These algorithms will provide Cyberworks Robotics a platform to build a new generation of industrial robot guidance systems.