Authors - Zain Nadeem, Hamza Nadeem, Jameel Ahmed Khan, Anayat Ullah Abstract - Indoor mobile robots have gone hand-in-hand with the automation of factories, households, and commercial spaces. They are present in all shapes and sizes, from humanoids used as waiters and personal assistants, to box-like warehouse item-sorters. Working of these robots depends on several key elements like localization, mapping, and surroundings detection. They also need to accurately detect walls and measure the distance to them, to be able to avoid collisions. Currently, this task is accomplished using LiDARs and other optical sensors which are costly, this can be catered for by using a vision-based solution. The main hurdle for vision-based systems is that walls do not have any distinguishable visual features to detect them. This paper proposes a system for measuring the distance to the wall by detecting the wall-floor edge. BRISK key-points are extracted on the detected edge and pixel-counting is then used to calculate the distance. The accuracy of the calculated distance is 95.58% and serves as a positive motivation for further development and refining of the proposed system. This paper implements the system on a single image which can be expanded to incorporate support for implementation on live video stream in the future.
Open Zoom