Research

In this work, we developed a scalable framework for early health monitoring in diverse care settings, leveraging four key vital signs—heart rate, respiratory rate, temperature, and oxygen saturation—that can be captured through commonly available means or unobtrusive wearable sensors. The framework integrates clinical criteria with advanced AI techniques, including ensemble modeling and an adapted BERT-based clustering approach, to classify individuals into health risk categories for timely intervention. Validated on publicly available PhysioNet CinC and MIMIC-III datasets, the framework demonstrates consistent and robust performance.  The real-time interface provides actionable insights for caregivers, enabling timely interventions and addressing resource constraints.

LiLoc is a system for precise 3D localization and tracking of mobile IoT devices (e.g., robots) in indoor environments using multi-perspective LiDAR sensing. The key differentiators in our work are: (a) First, unlike traditional localization approaches, our approach is robust to dynamically changing environment conditions (e.g., varying crowd levels, object placement/layout changes); (b) Second, unlike prior work on visual and 3D SLAM, LiLoc is not dependent on a pre-built static map of the environment and instead works by utilizing dynamically updated point clouds captured from both infrastructural-mounted LiDARs and LiDARs equipped on individual mobile IoT devices. To achieve fine-grained, near realtime location tracking, it employs complex 3D ‘global’ registration among the two point clouds only intermittently to obtain robust spot location estimates and further augments it with repeated simpler ‘local’ registrations to update the trajectory of IoT device continuously.

ERICA [paper] is a low-cost, pervasive digital personal trainer for users performing free weights exercises, with two key differentiators: (a) First, unlike prior approaches that either require multiple on-body wearables or specialized infrastructural sensing, ERICA uses a single in-ear “earable" device (piggybacking on a form factor routinely used by millions of gym-goers) and a simple inertial sensor mounted on each weight equipment; (b) Second, unlike prior work that focuses primarily on quantifying a workout, ERICA additionally identifies a variety of fine-grained exercising mistakes and delivers real-time, in-situ corrective instructions.

W8-Scope [paper] uses a simple, cost-effective sensor, containing only a 3-axis accelerometer and a 3-axis magnetometer, mounted on the weight stack of gym exercise machines, to obtain fine-grained insights into multiple aspects of gym exercise behavior. Using a multi-stage classification pipeline, W8-Scope uses the sensor to identify who is performing the exercise, what exercise they are doing, how much weight they are lifting, and whether they are doing the exercise correctly.  Compared to other solutions that require more extensive instrumentation or wearable devices, W8-Scope offers a compelling, low-cost, privacy-sensitive technique for such accurate monitoring. 

The IRIS platform [paper] uses a combination of smartphone and smartwatch sensor data to build a shopper’s profile based on inferring a shopper’s micro-gestural activities, such as “picking up an item” or “placing it in a shopping cart.” IRIS  uses standard locomotive and gestural micro-activities as building blocks to define novel composite features that help classify different facets of a shopper’s interaction/experience with individual items, as well as attributes of the overall shopping episode or the store. IRIS operates without any assumption of in-store infrastructure support or location tracking capability (no Wi-Fi, no RFID, no knowledge of store layout, etc.).

In [paper], we present the preliminary work on design and assessment of a game-based prosthesis training tool for upper-limb transradial amputees (amputations below the elbow). We re-designed open source implementations of two simple games and made them to be playable using three muscle contractions which are appropriate to pre-prosthesis exercises and are detected by an EMG-based arm sleeve. This training tool incorporates engaging and challenging elements that improve intrinsic motivation of subjects and also tracks quantitative outcomes such as muscle fatigue and muscle isolation of players during the game.