The purpose of my senior project is to design and build an automated tour guide system for the Union College campus. Currently, the admissions office offers hourly guided tours on weekdays from ten until three. Prospective students visiting campus at any other time are offered a self-guided tour, which consists of them being handed a campus map and exploring the campus on their own. As one might expect, these self-guided tours are not nearly as effective as the guided versions. This system is therefore intended as a replacement for the self-guided tours, and will provide a means for prospective students to have an informative campus tour regardless of when they choose to visit.
In order to be a useful tool, the system must fulfill a number of design goals. First and foremost, the system must be able to determine the user's location to an accuracy of five meters within selected buildings and ten meters anywhere outdoors on campus. Clearly, raw location information is useless to the user, so the system must also provide a visual representation of the current location. Additionally, the system must be able to determine routing information in order to give the user directions to a chosen destination. Finally, the system must provide contextual information to the user about nearby buildings/landmarks.
The main technical challenge associated with this project is determining the user's location, a process known as localization. In order for the system to function as desired, the user's location must be resolved in real-time, in three-dimensions, and both indoors and outdoors. Each of these constraints adds an additional challenge to the design.
A number of different localization technologies are being considered for this project. The most obvious and well-known technology is the Global Position System (GPS). This system uses signals from an array of geosynchronous satellites and a process known as trilateration to calculate a user's location anywhere in the world with an accuracy of approximately 10 meters. However, the signals transmitted by the satellite are too weak to penetrate most buildings, meaning that the system does not function indoors.
Two different technologies are being considered specifically for indoor localization. The Global System for Mobile Communications (GSM) is a cellular telephone communication standard that is fairly ubiquitous in North America, especially in urban areas. Recent research has shown that GSM can be used for fairly accurate indoor localization. Unfortunately, the relatively simple trilateration process used for GPS localization is not feasible with GSM, since determining the propagation of GSM signals is extremely difficult. Instead, GSM localization will be accomplished through a training process known as fingerprinting. The area of interest will be divided up into a grid of cells approximately one to two meters on a side, and in each cell, the signal strength of all nearby GSM cell towers will be measured. The combination of these readings constitutes the training set. In order to later determine a user's location, the same signal strength readings are taken. By comparing these readings to the training set, the most likely current location can be estimated using any of a number of different algorithms.
Another possible localization technology is the 802.11 wireless networking standard, commonly referred to as Wireless Ethernet. 802.11 localization can be accomplished in much the same manner as GSM localization, with 802.11 base stations taking the place of GSM towers. The much shorter range of 802.11 base stations means that a much greater density of base stations is required to achieve comparable results to GSM. At the same time, receiving a signal from an 802.11 base station provides greater location granularity than receiving a signal from a GSM tower, since the former restricts the user's location to a few hundred square meters, while the latter restricts the user's location to an area in the range of a square mile.
The localization technology employed in this design will be a combination of GPS and either GSM or 802.11. The use of GPS alone would provide accurate outdoor coverage but no indoor coverage, while the use of GSM or 802.11 alone would provide accurate indoor and outdoor coverage at the expense of extensive training. The combination of the two types of technology allows for maximum coverage with minimal training. The choice between GSM and 802.11 will most likely be determined by the density of GSM towers around the Union campus. Unfortunately, this information is not publicly available due to security concerns, and will therefore only be known after testing with a GSM modem.