High-Precision Vision Localization System for Autonomous Guided Vehicles in Dusty Industrial Environments

Abstract

Autonomous guided vehicles (AGVs) are widely used in ports and warehouses during the ongoing Fourth Industrial Revolution (Industry 4.0). AGVs can transfer cargoes to their correct destinations or carry industrial robots to work from various directions. In cases like Arthur et al. (2008) andHofacker (1993), engineers put a robot on an AGV to finish aircraft manufacturing instead of a fixed robotic system such as a gantry crane or guide rail on the ground. An AGV can avoid structural transformation and has high adaptability in various applications. However, highly accurate AGV localization is required for safe driving and precise robotic operations.Several studies have investigated high-precision localization methods for AGVs since 1908. A classical method involves setting magnetic or electric cables underground for tracking. AGVs would follow the cable path by electromagnetic induction (Chen et al., 2004) using special sensors. This method is still widely used in modern factories. It can maintain a position accuracy of around 10 mm. However, it damages the original ground and can be easily affected by magnetic and electric fields.Some studies used a monocular camera (Wu et al., 2019) to track colored stripes and quick response (QR) codes. The positioning error in those studies could be

An accurate and robust visual-inertial positioning method

“…Hijikata et al presented a simple self-localization method that used a single camera observing several LEDs set at known positions [21]. Liu et al developed a vision localization system with LED array targets and a camera installed on a rotating platform for autonomous guided vehicle (AGV) navigation [22]. In order to better meet the needs of the human-machine integrated coordinate measurement, an accurate positioning method using a multi-camera system (MCS) with non-overlapping views was previously proposed [23].…”