Designing a ubicomp system for neighborhood sustainability
Have you ever walked around neighborhoods in a city like Philadelphia and noticed a large variety in cleanliness from block to block? What about scrolling Facebook or Nextdoor and seeing ranting posts from a neighbor directed towards an unknown perpetrator or group or perpetrators of littering? What if there was a way to leverage ubiquitous computing to lessen these issues and live in a cleaner and more environmentally sustainable city?
Neighborhood Sustain is a system that seeks to improve awareness and initiatives for handling cleanliness and environmental sustainability, especially in urban neighborhoods with high traffic. It combines neighborhood-level data with city and state data to quantify and track progress. Data is collected from distributed data sources, including Nextdoor, Facebook groups, Ring and IP Camera Feeds, Google Street View, user submissions via the web app, maintained neighborhood IoT sensors, as well as any applicable city or state data sources.
First, let’s define and separate the concepts of cleanliness and environmental sustainability, the two key issues that Neighborhood Sustain seeks to address. Cleanliness is a more demanding, day to day goal for a neighborhood. It involves considering sources of trash and waste as well as the disposal outlets for these. Illegal dumping, windy trash days, full receptacles, and littering are common causes of cleanliness issues. Sustainability in the context of Neighborhood Sustain is a larger, longer-term goal. It not only considers historical cleanliness data, but also considers air pollution, utility usage, waste production, transportation, and more. It seeks to create longer-term goals and objectives for a specific neighborhood.
The system behind Neighborhood Sustain’s cleanliness system (“Clean”), will be the most widely used as it will deliver faster, more noticeable, and actionable results than the sustainability offering (“Sustain”). Clean’s backend will be cloud-hosted and will consist of a machine-learning/AI implementation tracking active issues around cleanliness. The Ring/IP Cam data source will consist of live outdoor camera feeds installed and maintained by residents. These feeds will stream to the main system, which will use image recognition to identify trash in the neighborhood such as loose papers, glasses, bottles, dog poop bags, and more. This will be tallied into an active count of “active” litter in the neighborhood. Optionally, the system will track the source of the litter if from an individual if enabled by the feed owner/specific neighborhood. In addition to image-recognition powered camera feeds, Clean will also periodically check neighborhood Facebook and Nextdoor Groups via an API for posts calling out littering or trash. These occurrences will be added as separate incidents if not covered by a camera feed and a higher severity weighting will be placed upon it by the classification algorithm. Additionally, part of the web interface will allow for user reporting of trash and incidents. Also, IoT sensors will be deployed in the neighborhood to track trash production and litter. A pair of waterproof IoT sensors will be attached to trash receptacles — for weight and capacity respectively to prioritize receptacles to be emptied. Additionally, city and other public data sources will be leveraged to receive/transmit data from the Clean system. Receptacles to be emptied or active incidents will be escalated directly to sanitation departments or municipal services for awareness, and weather data will be leveraged for wind and precipitation information for awareness on trash days. If a threshold has been reached, the system will trigger a city block street sweeping response.
User usability is a major component of consideration when proposing the Clean system design. After all, why not just stick with traditional methods such as organizing volunteer groups or block captains versus the cost and legwork of building this system? The benefits are countless and will lead to greater education and progress towards sustainability goals when combined with the power of modern computing technology. Instead of simply having a reactionary response to neighborhood cleanliness such as weekly clean-ups, Clean seeks to educate and resolve problems at the source. All residents will be given the option to join the Clean system and create an account. Residents will have the option to receive push alerts for detected litter and by picking it up and disposing of it can receive credit (tallied on a running leaderboard). Additionally, Clean will track each residence’s trash days and waste management system. If their trash is successfully intact right before collection, they will receive a bonus. On the contrary, if the system detects loose trash before collection, they will be negatively ranked in the system. Having this interactive leaderboard seeks to inspire a sense of community yet competition among residents to keep their neighborhood clean.
Sustain is the longer-term environmental sustainability service from Neighborhood Sustain. As mentioned earlier, Neighborhood Sustain will start its model’s basis by tracking historical cleanliness data from Clean. Neighborhood Sustain will rank a neighborhood’s sustainable index from a multitude of factors, as well as potential actionable improvements to improve this score. Neighborhood Sustain will again use image recognition from Clean camera feeds for monitoring historical vehicular traffic and quantifying neighborhood waste production by volume. The system will weigh pedestrian, bike, passenger car, and bus traffic for its mobility calculation. Additionally, Sustain will connect to IoT sensors around the neighborhood such as air pollution monitoring and UV/ambient light sensors. Localized, historical air quality monitoring data such as the Urban Air project from Microsoft Research in Chicago could feed the Sustain models. Finally, the last initial implementation to the Sustain system would be with public utility companies and city and state governments. Sustain would classify the cleanliness of a neighborhood’s energy suppliers from publicly available data and would prompt residents to track their historical usage. Sustain will also have a leaderboard component but objectives will be much higher-level and longer term. Based on the analysis of historical data, recommendations for improving sustainability could be presented to community leaders and local governments. Community effort and property values could theoretically increase with a high sustainability index, bringing more wealth to the local community. There is an education component that will be tied into Sustain to teach youth about how to make their communities a more environmentally sustainable place.
Neighborhood Sustain’s pricing structure is expected to be offered in multiple tiers depending on neighborhood size as well as integrations and support. Neighborhood Sustain could partner directly with HOAs and other private communities. In addition, Neighborhood Sustain will have a reduced cost subscription option partially funded by city and state governments in high-need areas. This is where incentivization and education opportunities could come into play. Finally, Neighborhood Sustain could potentially establish government and private contracts for corporate campuses and public areas.
Neighborhood Sustain is a project that has a wealth of benefits and potential for urban and suburban neighborhoods alike. By harnessing the power of machine learning/AI and cloud computing, technology can create direct actionable tasks for residents and generate intelligent insights. The ability to provide real-time trash management capabilities and longer-term sustainability planning via technology will result in direct positive community development. Neighborhood Sustain has the potential to make Earth a more habitable place to live now and in the future.
