PISALA project. Intelligent Sensorization for Line tracking with Artificial Vision

This paper presents an artificial vision embedded solution for detecting painted lines on the floor, in the context of the Research Project PISALA (Industrial Prototype for Automatic Line Tracking with Artificial Vision), whose main objective is to provide vision-based solutions for Automated Guided Vehicles (AGVs). In addition the paper proposes a procedure for evaluating robustness and accuracy of line-detection algorithms including qualitative and quantitative tests for detecting specific situations such as images with multiple lines or lines with spurious objects, lines with different colors, etc. The paper also considers the algorithm's sensibility with respect to extrinsic camera calibration errors, in order to define the bounds of the detection area. As a result, low cost robust algorithms for line detection can be evaluated. The examples for validation have been implemented on the embedded vision system CMUCAM3.