The Projects

Last Year’s Event

Last year we held an inaugural Pilot Pitchfest, inviting New York City agencies and researchers to deliver low-stakes, two-minute pitches on pilot projects they hoped to run in the following 12 months. Some stuff worked well. Some stuff didn’t—and we learned a lot in the process. In response to popular demand, in Year II, we expanded the program to include all city agencies and academic disciplines, and both research and technical assistance project tracks. 

Download the Deck  Watch the Recording

Funded Projects

Following the Pitchfest, agencies and researchers were able to apply for funding. Two projects were selected and launched in Summer 2024.

Using Computer Vision to Automate Bike Lane Inspection

The Context: The New York City Department of Transportation (NYCDOT) conducts regular assessments of pavement conditions in roadways, utilizing a manual inspection process in which NYCDOT staff drive, and simultaneously score, roadway conditions throughout the city. The inspection process has become more complex due to an additional layer of urban infrastructure introduced over the past two decades: the implementation of over 250 miles of protected bike lanes on city streets. As NYCDOT’s inspectors’ views of the lanes are blocked by parked cars or concrete barriers, the lanes cannot be covered by  NYCDOT’s existing inspection processes. Simultaneously, in the years since NYCDOT first developed its roadway inspection process, there have been significant developments in computer vision technology. NYCDOT hypothesized that there may be a way to automate the inspection of pavement conditions in protected bike lanes using computer vision—and that the learnings from a controlled experiment in bike lanes could potentially also inform how they inspect regular roadways. 

The Project:  NYU developed, deployed, and refined a novel vision-based deep learning model designed to automatically assess pavement conditions in protected bike lanes. Through training, the model was able to address the unique challenges of automated bike lane pavement assessment, including the need to detect both pavement distress and frequently occurring street hardware (e.g., drains and grates). Field testing and validation were conducted on 22 street segments,  with protected bike lanes in New York City using both a regular bike and a cargo bike, with an attached BSafe-360 bike sensor built by the research team, integrating cameras, GPS, and accelerometers to collect synchronized data. The detection outputs, combined with accelerometer data, were used to generate a Bike Lane Defect Index (BLDI), which is a composite index that assesses the number, type, and extent of identified defects, as well as surface unevenness for each bike lane segment. The results demonstrate a vision-based deep learning solutions can offer sufficient practical performance to accurately identify bike lane defects in real-world scenarios. While the training of the model and the development of the BLDI has inherent value, further efforts are still necessary to create a commercial product that can actually be scaled across DOT’s inspection processes.

Impacts

  • Agency: NYCDOT is exploring issuing a procurement for a commercial solution that leverages computer vision for roadway inspection, informed by the results of the pilot project and other adjacent initiatives to leverage computer vision across DOT’s portfolio.

  • Academic: NYU presented the project at the Transportation Research Board Annual Conference. The team intends to publish the work in a research journal and summarize the lessons learned for the U.S.DOT ITSJPO's ITS deployment evaluation database.

Principal Investigators: New York University  

Collaborator: New York City Department of Transportation (NYC DOT)

Validating the Impacts of Leaf Mulching on Urban Soil Health

The Context: Every fall, NYCHA staff spends significant time raking leaves from over 1,000 acres of tree canopy and transporting those fallen leaves offsite. NYCHA’s Sustainability and Waste Management Departments were in the second year of a pilot in which, instead of raking, staff used retrofitted mowers to mulch leaves directly onto grass. There is an upfront cost associated with procuring and installing mulching attachments for mowers, which NYCHA needed to justify to expand the program from pilot to roll-out across NYCHA's building portfolio. Leaf mulching was already known to save staff time, sequester carbon and reduce carbon emission, and NYCHA hypothesized it also improved urban soil health, across variables like total organic content, infiltration capacity and microbial mineralization rates. However, NYCHA struggled to find scientific data to substantiate the soil health benefits of leaf mulching that it could use to make a case for investing in an expansion of its leaf mulching program.

The Project:  Dr. Zhongqi (Joshua) Cheng—a professor in the Department of Earth and Environmental Sciences at Brooklyn College—led a study to validate the soil health improvements of leaf mulching for NYCHA lawns. Dr. Cheng hired a student research team, who collected and analyzed soil samples from five different NYCHA sites across three boroughs, coordinating with NYCHA groundskeepers. Samples were analyzed to determine the program’s impacts on a range of soil physical, chemical and biological parameters, such as bulk density, water holding capacity, pH, organic content, nutrient content, microbial biomass and respiration. After determining the initial sample set was insufficient to discern a clear pattern — likely a result of the fact that mulching had only occurred for one season — the research team then expanded its sampling to include two additional sites run by the New York City Parks Department, where leaves were not raked and left onsite for multiple seasons. Through this larger sample size, the research team was able to deduce a positive impact from leaf mulching across most soil health properties. However, more longitudinal data is required to establish a definitive pattern.   

Impacts

  • Agency: NYCHA received a grant from the U.S. Department of Agriculture that includes funding to expand its leaf mulching program, as well as soil sampling (of leaf composting and compost generated at NYCHA) that will be sent to Dr. Cheng’s lab at Brooklyn College. NYCHA and Brooklyn College are now exploring the possibility to establish a City-wide soil testing program, to track longitudinal data related to leaf composting and mulching over time.

  • Academic: In addition to one graduate and one undergraduate student that this pilot funding supported, Brooklyn College was able to secure NSF REU funding to support three undergraduate students on the project, extending the impact of program dollars. Brooklyn College is submitting its research on the efficacy of leaf mulching to a peer reviewed high impact journal. 

Principal Investigator: Brooklyn College 

  • Zhongqi (Joshua) Cheng, Professor of Environmental Geochemistry, Urban Soils and Urban Sustainability, Director of Environmental Sciences Analytical Center

Collaborator: New York City Housing Authority (NYCHA)