Indoor Augmented Reality Research

Projects > Indoor Augmented Reality

ABSTRACT

Location tracking is an important part of many ubiquitous computing applications. Unfortunately, global positioning systems (GPS) only provide sufficient tracking coverage in an outdoor setting, unavailable to many users in an indoor setting. With the popularity of wireless access points, and WiFi-capable mobile devices, the use of this wireless infrastructure can become a viable option for instant and portable indoor tracking. In this paper, we investigate and purpose deploying a quick real-time indoor location system that accounts for obstacles and wireless instability for mobile devices without fingerprinting locations. We discuss the actual performance and accuracy of tracking both stationary and moving targets in a structure by incorporating common mobile device tracking features, and applying predictive heuristics to account for inaccuracies such as structural boundaries. Provided with only a passive based collection of WiFi based signal strength indications (RSSI), this system relies on achieving accuracy by using the built-in orientation features present in many existing mobile devices, and does not rely on any costly additional sensor/external feedback. By using the collected signal graph, a positioning map is formulated and used with both predictive and deterministic algorithms to derive the estimated location of the WiFi node. We compare the performance of the algorithm with several other approaches, including fingerprinting approaches. Our real life walkthroughs collecting data samples indicate that once a indoor structure is passively scanned, our system is sufficient enough to derive accurate location estimates of nodes with decent accuracy. With such information easily obtainable, a scalable context-aware mobility framework was developed to demonstrate a context-dependent multimedia interactive application.

It is often the case that indoor localization trackers are unaware of the access points. Rather then construct a model of the existing wave propagation and tracing the source, develop a model that follows the existing signal trend per access point and takes into account both new mobile tracking technologies (accelerometers, GPS), and signal inconsistencies that would question the accuracies of other wireless models. Tracking targets in real-time involves complexities such as moving obstructions, variations in readings, and other signal discrepancies.

Accuracy

• Predictive location algorithm (given obstacles, direction, and scan information).

Convenience

• Passive (non-automatic) collection of flagged location information.

• Requires no knowledge about the source of the access points.

Portability

• Only uses publicly available access point information.

• Deployable on mobile handheld devices.

Affordability

• Uses existing established WLAN/WI-FI standards.

• Ability to run on any mobile platform without additional sensor feedback.

We're currently developing a context-aware mobility framework that supports context-dependent, multimedia interactive application (e.g. augmented reality viewer).

If you have questions or thoughts, please contact us here.