A wireless infrared sensor network for the estimation of the position and orientation of a moving target Nikos Petrellis, Nikos Konofaos and George Alexiou Computer Engineering and Informatics Dept. University of Patras Rio Campus, Patras, GREECE +302610996933 petrelis@ceid.upatras.gr ABSTRACT The location of a moving person or vehicle in a virtual reality environment is a critical issue. A wireless infrared sensor network capable of estimating the position of a target on a plane and its orientation is presented in this paper. This is actually an extension of the position estimation system that was presented in [1-2] where the target was allowed to move on a plane but not to rotate. In [1-2] the estimation of the position was based on a network consisting of a few low cost infrared transmitters and a pair of infrared receivers mounted on the target. In this paper we discuss how a third infrared sensor properly positioned at the side of the receiver can also allow the estimation of the target orientation. Keywords Position localization, infrared sensors, wireless sensor networks. 1. INTRODUCTION Several approaches that have been proposed in the past, for target localization in automation and robotics are based on different types of media. The dominant methods rely on measuring the RF or ultrasound signal strength of multiple transmitting devices [3] or the time of the fly of short (laser, infrared) signal pulses reflected by the surface of the object under measurement [4-7]. Both of these approaches require high quality analog sensors and fast processing units capable of measuring accurately very short time intervals. In robotics and virtual reality applications this procedure is assisted by additional methods like image processing, odometers, accelerometers etc [8-10]. The architectural and computational cost of these approaches is significant. We have recently exploited the quality of the received infrared patterns in order to estimate the distance of the receiver from the transmitter. More specifically, we have shown [2] that the position of a moving target on a grid plane of 15m 2 can be estimated if two infrared pattern transmitters are positioned at the borders of the covered area and a pair of sensors is mounted in opposite directions at the side of the target. The area covered can be extended if multiple transmitters are used. Nevertheless, it is essential that the target does not rotate in order to predict its position accurately. In the present work, we discuss the way a third sensor at the side of the receiver can resolve the aforementioned limitation. The position estimation history is also used in order to increase the precision of the estimation and extend the covered area when multiple infrared pattern transmitters are used. In Section 2 we briefly describe the architecture of the system described in [1-2] and present the extensions added in the current work. In Section 3 we explain the position and orientation estimation method that is demonstrated by a simple case study. 2. SYSTEM ARCHITECTURE The position localization system presented in [1-2] consists of the infrared transmitting devices IRTX1 and IRTX2 that are placed around the covered area as shown in Fig. 1. These devices transmit different sets of patterns that are received with varying quality by the pair of receivers IRRX A and IRRX B at the side of the target. The reception quality is defined as the number of the received patterns of a specific type compared to the expected ones in a specific time interval (success rate of this pattern). The system cost is drastically reduced using this method since we merely count digital patterns instead of analog signal strength or very short time intervals. Each one of the transmitting devices supports a set R of pattern types (R1 for IRTX1 and R2 for IRTX2). There are no common pattern types in these sets (R1R2=Ø). The pattern types differ in the number of pulses they consist of and their duration. A pattern type MODi consists of i pulses. If i>j then MODj will be received with a higher success rate than MODi at the same position since it consists of smaller number of pulses (there is higher probability to receive a pulse distorted in MODi than MODj). The pulse durations of MODi are defined shorter than those of MODj in order to further strengthen the previous assumption. Before real time operation, the target visits the nodes of a virtual grid that covers the area around which the target moves and records the success rates of the various patterns received (calibration stage). The set of these rates define a unique identity for each grid node on the plane. During real time operation the retrieved success rates at the current position are compared to the identities stored from the calibration stage and the closest node is selected. Then, an interpolation search in two dimensions is Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Mobimedia'07, Month 8, 2007, Nafpaktos, Aitolokarnania, Greece. Copyright 2007 ICST 978-963-06-2670-5.