Cities across the globe are undergoing rapid changes and growth. They are becoming smarter, not just in the technology used, but how city planners design the spaces where we live, work, and play.
In order to properly plan a city, the rules that govern it, and the ability to enforce the laws, city managers need the right tools and data.
productOps LOUD “Live Outdoor Urban Data” tackles the problems that cities face around the noise we create and how it impacts others. For over 100 years city noise issues have largely been seen as a division between economic areas in a city. More affluent neighborhoods have historically had more noise ordinances and higher rates of violation enforcement.
LOUD provides a way to measure city noise and make that data available to the public. It records outside noise levels in perceived decibels and identifies the source of the noise (music, construction, vehicles, industrial equipment, etc.)
In addition to measuring and classifying city noise, LOUD can respond to violation events in real time by using VANTIQ collaborations. Administrators can create rules in VANTIQ that identify a violation and then create collaboration tickets. The VANTIQ client app also allows citizens to report tickets helping make the automated machine learning smarter.
HOW IT WORKS
LOUD has four major components:
1) Edge devices that record sound and measurements
2) VANTIQ Modelo platform to process noise and manage workflows
3) AWS SageMaker machine learning to classify noise
4) VANTIQ Client to create dashboards and mobile applications for users to interact with the data and collaborate in real-time.
We are using a series of audio sensors to measure noise levels at various locations. Each location has two sensors, one to measure the inside noise level and one for the external noise level. This helps determine the source of the noise by comparing the two, especially in locations that are partly open to the outside or that have poor noise isolation. In addition, the outside sensor also records the actual sound and saves it in a special AWS S3 bucket used by AWS SageMaker.
The Edge Devices send their data to the VANTIQ platform. This updates VANTIQ with real-time updates for each location as well as storing historic information for each sensor. Applications in VANTIQ determine if the audio event is above the noise threshold and if so, sends that data to AWS SageMaker for ML classification. If the classification, duration, and noise level meet the criteria of a violation, then a Violation Event occurs and triggers a VANTIQ collaboration.
LOUD utilizes SageMaker to process the audio recordings and determine what the predominant noise is in the recording. This information is then sent back to VANTIQ. In order to have accurate classification, productOps created training models based on crowd sourced pre-recorded city noises. The models are continually updated and learn as more data is processed.
There are two VANTIQ clients that create real-time data visualizations and user interactions.
The first is the Administrator Dashboard that allows a manager to see what the current noise levels and classifications are at each location. Users can view a map of the city with updated information or drill down to a specific location to get more details and historic data and trends. It also lists recent violations and their status.
There is also a mobile app that uses VANTIQs mobile client to allow city managers and citizens to collaborate on creating and responding to violations.