Optimum landmark placement for improving accuracy of pedestriandead reckoning in indoor localization

Localization is a technical to find the location of a target, such as a worker walking in a plant. For outdoor localization, satellite-based global positioning system (GPS) is commonly used. GPS does not provide good accuracy for indoor localization due to the complicated indoor environment that affects the GPS signal propagation. Currently, there are two types of indoor localization methods, one is based on existing indoor wireless infrastructure (such as WiFi), and another is based on data collected by motion sensors that are attached to the target object. Both methods do not work well in many industrial environments. The WiFi coverage is either incomplete or unstable due to physical limitation or strong interference caused by machines; while algorithms using data from motion sensors can cumulate errors that exceed a tolerable level within short distances. Combining both methods helps improve the localization accuracy, but still requires sufficient WiFi coverage. Placing landmarks at fixed and known locations helps recalibrate the latter method by resetting the error to zero or near zero. In this research, we will study how to minimize the cost to place the landmarks, while achieving the required localization accuracy.

Faculty Supervisor:

Dongmei Zhao

Student:

YiQiong Miao

Partner:

Muldi Vision Ltd.

Discipline:

Engineering - computer / electrical

Sector:

Professional, scientific and technical services

University:

McMaster University