WLIA 2020 Spring Virtual Meeting

In March 2019 the Minnesota Pollution Control Agency slapped a historic $7 million penalty on the Water Gremlin company; for releasing excessive amounts of carcinogenic industrial solvents into the air. For 17 years, surrounding neighborhoods were unknowingly exposed to a toxic airborne substance at levels 12 times higher than considered safe. With air quality sensors now relatively common in highly urbanized areas; how could this go unnoticed for so long?

Nearly 2.5 billion people globally have Internet-connected phones. The computational capabilities of these devices render them extremely effective at real-time edge computing; by communicating directly with nearby sensors and performing onboard data analysis.

In this talk we’ll discuss an approach whereby human sensors collect discrete observations with the Fulcrum field data collection app. Simultaneously, the device communicates with nearby sensors; Flow by Plume Labs measuring air quality, SensorLog measuring sound pollution and Campbell Scientific devices monitoring transportation infrastructure. Analytics are performed on the device, combing human observations and raw sensor data into immediately actionable insights.

We’ll show how this approach is applied using three real-world scenarios; smart cities and healthy environments, transportation infrastructure and mining operations. We’ll demonstrate the value of context-driven data collection and situational human judgement which cannot (as of yet) be replaced. We will wrap up our discussion with a few thoughts on TensorFlow Lite, an open source deep learning framework for on-device inference.

https://attendee.gotowebinar.com/register/2551954800481330192