Two Proposed Indoor Multi-Cameras Positioning Systems Compared to Classical Geometry System

Positioning systems in indoor environments are of a great concern in automation and robotics domains where performing critical tasks requires precision. However, to make these systems widely applicable they must be cost-effective. The objective of this paper is to develop two different 3D positioning systems based on neural networks and adaptive neuro-fuzzy techniques. Sample images of a recognizable object were taken using three low-cost cameras as training and testing data for these systems. Positioning results of the proposed systems are compared with results of the classical geometrical method. The results show positioning errors on the scale of millimeters and the neural network system produces the smallest error.