This paper proposes and presents the first experimental demonstration of a high-precision indoor 2D and 3D visible light positioning (VLP) system using an imaging multiple-input multiple-output (MIMO) configuration with supervised artificial neural network (ANN). The proposed system utilizes four distributed transmitters and receivers with four photodiodes and an imaging optics. The experiments are conducted in a typical indoor environment with
transmitter separations of 300 mm and a link distance of 1400 mm. The experimental results show 2D and 3D positioning accuracies of 3.7 mm and 51 mm, respectively. A simulation model is also developed for the VLP system
to verify the experimental results. Further optimization of the VLP system in the simulation platform leads to improved 2D and 3D positioning accuracies of 2 mm and 9.3 mm, respectively. The proposed system can seamlessly converge
with existing lighting infrastructures and is also compatible with the imaging MIMO visible light communication (VLC) system, indicating the potential for practical implementation in integrated communications and positioning applications