Design, Calibration, and Evaluation of a Long-Range 3-D Infrared Positioning System Based on Encoding Techniques

Abstract

Optical indoor positioning systems have experienced an increasing research interest during the last decade as they can provide 3-D centimeter accuracy using light emitting diode (LED) lighting. In the case of having several LEDs emitting simultaneously and a receiver (e.g., the tag to be localized), these systems face important challenges, such as very high dynamic ranges with low signal-to-noise ratios when the coverage area is increased, multiple access interference (MAI), multipath and near-far effects, calibration issues (misalignments in the receiver and other intrinsic parameters), and so on. Previous work have already shown the feasibility of using LED emitters in combination with quadrature angular diversity aperture (QADA) receivers to implement positioning systems. This work further introduces new design considerations, tested on an experimental setup, to enlarge the emitter–receiver range, thus increasing the total coverage, while dealing with the aforementioned challenges. The system applies encoding techniques to each transmitter to solve the multiple access problem. The performance of two different types of codes has been compared, as well as their influence on the final estimation of the receiver’s position: one based on Kasami sequences and another based on loosely synchronous (LS) codes, derived from complementary sets of sequences. At the receiver, the estimation of the incident point is constrained to angles at which the system can be linearized, and a specific calibration process for this type of sensor has also been defined and applied. The proposal has been finally validated with simulated and experimental results in a large space of $2 \times 2$ m2 (base), with a distance from transmitters to receiver of 3.4 m (height). The experimental tests at distances up to 4 m, carried out after the calibration process, achieve average absolute errors below 10 cm for $X$ and $Y$ axes and around 20 cm for $Z$ axis, and standard deviations below 4 cm for $X$ and $Y$ axes, and around 30 cm for $Z$ axis.