AWS-Based IoT Data Aggregation and Analytics Platform

It is entirely API-driven and tremendously scalable and extensible. Through an open API architecture, data can be streamed from any wearable device or machine, including belts, vests, and hard hats. The connected workplace can be extended to phones, chip-enabled ID badges, forklifts, trucks, and factory machinery. Organizations can also move the data to their own environments for advanced analytics and third-party device integrations.

The data is securely received in millions of bits and is intelligently reassembled. The back-end application platform provides an advanced framework for a company to monitor data intelligently and receive alerts based upon established thresholds built using machine learning algorithms.

Data is parsed through pattern recognition and data modeling and is then converted into simple verbs, think walking, driving, bending, climbing, and sitting or standing (idle). The SmartBelt platform employs two random forest models within the batch processing. The first model is a high-level activity classifier and provides an activity label for each millisecond of data (e.g., static, dynamic, upper-body, or lower-body movement). This classification is included as an input (feature) in the second model, which provides a lower-level classification for each second of movement (e.g., walking, sitting, or twisting).

Beyond this, the platform employs many heuristics to arrive at other classifications like driving, acceleration, hard braking, and squatting that lead to intelligent Gait analysis'more specifically, the study of human motion-using the SmartBelt to augment the instrumentation for measuring body movements, body mechanics, and physical activity at a particular geo-location.