Indoor Localization

Participated in improving indoor localization with Bluetooth Low Energy (BLE) and Wi-Fi RTT beacons. Created map visualization to demonstrate the quality of the localization and to find approaches for improvements. Incorporated the technology into applications such as optimizing parts deliveries in a factory.

Technical challenges. Fitted BLE path loss model and Wi-Fi RTT distance measurements to ground truth. Implemented the particle filter framework in C#, Java, and Python. Developed map visualizations in Xamarin and Python Matplotlib. Conducted analyses with Numpy. Configured Wi-Fi RTT access points for best performance.

Technologies. Xamarin, .NET, C#, Android, Java, Python, Numpy, Matplotlib.

Fusing map information with sensor model

Video Surveillance

Created a video surveillance system that combined computer vision with intuitive user interfaces. Provided access to many recorded or live video streams to several clients in parallel. Used computer vision to locate people in a map and across cameras.

Technical challenges. Recorded video from more than 20 cameras as motion JPEG to support playback at variable speeds and in reverse. Handled more than 1 TB of daily data. Devised a system to smoothly record and stream that video. Designed and implemented a UI that provided access to the video as synchronized playback for all cameras with higher frame rates for video in focus. Used computer vision results provided by others to determine cameras of interest and to track people on a map. Created a spatio-temporal analysis and displayed it in the form of time-segmented heatmaps.

Technologies. Java (server and client), JPEG, REST.

Spatial arrangement of camera views