Niranjini Rajagopal
Carnegie Mellon University
niranjir@andrew.cmu.edu
Bio
Niranjini Rajagopal is a PhD candidate in electrical and computer engineering at CMU, advised by Anthony Rowe and Bruno Sinopoli. She received an master’s degree from CMU and a bachelor’s degree from the National Institute of Technology, Trichy, in India. Her research is at the intersection of embedded sensing systems, estimation theory, and statistical signal processing. She collaborates with industry and standards organizations, and has interned at Texas Instruments embedded processing systems lab and the wireless location group at Apple. Her work at Apple resulted in product impact. She has been part of the multi-university TerraSwarm and CONIX research centers. She has published in several ACM/IEEE conferences (IPSN, SenSys, RTAS, ICCPS, RTSS), won IPSN best demo and won the Microsoft Indoor Localization Competition. She received the CMU William S. Dietrich II Presidential PhD Fellowship in 2015 and the Samsung PhD Fellowship in 2016.
A Sensor Fusion Approach to Indoor Localization
A Sensor Fusion Approach to Indoor Localization
Being able to locate people and things inside a building accurately, instantly, and cost-effectively will revolutionize how we interact with our indoor surroundings and open up new application domains. Indoor localization at scale is not yet a reality due to technical barriers in terms of what sensors are available on commodity devices and gaps in scaling these systems from labs to realistic building environments. My work takes a principled approach to solving the problems faced by emerging ranging and localization technologies while scaling up with regard to sensor placement, mapping, and noisy sensor measurements. I apply tools from estimation theory and statistical signal processing to embedded sensing systems. My work addresses three major localization problems by fusing sensor data and floor plan information. First, we explore quantitatively how beacon placement impacts localization performance in complex indoor environments. Second, we design a floor-plan aware localization solving technique that can accurately localize in the presence of inaccurate non-line-of-sight signals and insufficient line-of-sight signals. Third, we design a crowdsourced pedestrian-based mapping system on mobile phones for rapidly configuring beacon infrastructures. We have demonstrated our techniques with multiple real-world systems. Our work with ultrasonic beacons and unmodified phones won the Microsoft Indoor Localization Competition in 2015 and spun out into a start-up. Applied to ultrawide-band radios, our system won the same competition in 2018 with the ability to support 3D tracking. We also built a persistent, multi-user augmented reality application on mobile devices, which won the IPSN Best Demo Award in 2018. We are now working with NIST on localizing firefighters during operation without dedicated infrastructure in the building.