Ambient data collection in indoor building environments using mobile robots

Building designers have increasingly looked at simulation to improve the performance of building systems with the primary objective of simultaneously maximizing comfort, and minimizing energy use. The quality of the simulations depends on the quality of the data being input into the models. The input data for simulation comprises the current real state of a building which encompasses several components that include the state of occupants, comfort parameters, and the building systems. Typically, such data is aggregated using preinstalled Building Automation Systems (BAS) with the help of stationary wired or wireless sensor networks. Such a process is cost-prohibitive, time consuming, and often impractical in existing buildings without BAS. This paper proposes mobile platform based robotic data collection for gathering energy and comfort related data in real-time which can be utilized for further simulation analysis and decision-making. The fiducial marker based navigation and drift correction algorithms developed to facilitate the robotic platform navigation in a building are discussed in detail. This method successfully achieves the navigation task by providing directional navigation information along with drift correction at critical discrete locations instead of the traditional continuous updating process, which is computationally intensive. An experimental study validating the statistical equivalence of the two data sets gathered by the traditional preinstalled fixed sensor networks and the multi-sensor fused robot was performed. The results demonstrate the feasibility of the proposed methodology in efficiently collecting large datasets in buildings using only a single set of sensors in contrast to the scads of similar sensors required by traditional data collection methods.