Designing Automated Guided Vehicle Using Image Sensor

Abstract

Automated guided vehicles (AGV) are one of the greatest achievements in the field of mobile robotics. Without continuous guidance from a human they navigate on desired path thus completing various tasks, e.g. fork lifting objects, towing, and product transportation inside manufacturing firm. Their development can revolutionize the world in the sense of fool proof navigation and accurate maneuvering. Though most of the presently the AGV work in a retrofitted environment, work space as they require some identification for tracing their guide path, works are going on developing such AGVs which are dynamic in sense of navigation and whose locomotion is not limited to just a retrofitted workspace. The aim of this work was developing such a natural feature AGV which takes visual input in the form images and gains detailed object, obstacle, landmark, identification to decide its guide path. The AGV set up developed, used a commercial electric motor based car ‘Reva i’, as chassis which was fitted with camera to take real time input and resolve it using segmentation and image processing techniques to reach a decision of driving controls. These controls were communicated, or better imparted to vehicle using parallel port of computer to servo motors, which in turn controlled the motion of vehicle. The work was focused more on dynamically controlling the vehicle using refinement of driving mechanism (hardware), however it could be assisted using better segmentation and obstacle detection algorithm. All the retro-fitting and codes were developed in such a way that they could be improved at any stage of time. The results could be enhanced if a better stereoscopic camera were used with a dedicated cpu with better graphics capability. This vision based AGV can revolutionize the mobile robotics world, including systems where a human driver is required to take decisions on the basis of visualized condition.