Planning & Environment

Drone-based hyperspectral imaging technique for monitoring plastic pollution

We often see plastic bags or other garbage hanging on trees, along the roadside, slipping down the storm drain, and floating in the ocean. There are over 51 billion pieces of litter on our nation’s roadways, 4.6 billion of which are larger than four inches in size. Plastics takes 19 % of 51.2 billion pieces of litter on roadways nationwide. 1.4 billion beverage containers on our nation’s roadways: Beer bottles: 30%; Soft drink: 25%; Water and sport drinks: 10%. Site attributes correlate with amount of litter. Residential areas were 40% less littered than roadways in general. However, roadways near convenience stores were 11% more littered. Roadways near commercial establishments were 11% more littered. Specifically, for traffic systems, plastic pollution also depends on road attributes, which is not well studied. Plastic pollution lead to clogging of storm drainage and overflow or flooding of storm water, and contamination. Plastic pollutants may obscure traffic signs and lights and impede roadways, which increases traffic jam and accidents due to decreased visibility and traffic control. 25,000 accidents are caused each year in the US by litter. Over 800 Americans die each year from crashed related to litter and debris. To effectively mitigate plastic pollution on roadways, implementing cost efficient monitoring and cleaning measures is critical.

 

Current practices are primarily manual and routine inspection of roadways, which could not efficiently detect accidental pollution and have insufficient coverage especially in remote or suburban roads. Sustaining consistent, accurate and long-term monitoring networks is important for traffic pollution control and mitigation. This project aims to explore a novel imaging and mapping technique, empowered by rapid detection algorithms and dynamic path planning algorithms on hyperspectral spectroscopy and unmanned aerial systems (or drones). The benefits of this advanced technology are expected to include the efficient and high-resolution imaging data acquisition on inspected roadways. Swift localized monitoring of plastic pollution on roadways will enable proactive measures or strategies for cleaning and correction, which prevent traffic problems, flooding or accidents.

 

Small unmanned aerial vehicles (sUAVs, a.k.a., Drone) have been used for a variety of purposes, including: law enforcement, landscape filming, commodity delivery, avalanche monitoring, and other civilian applications. Equally, drone systems hold great promise in traffic systems for monitoring. A sUAV is able to maintain a higher fuel to weight ratio due to the lack of weight and volume of not having a human crew and their supporting equipment. sUAVs generally have less capital cost (e.g. manufacture cost) and operational cost (e.g. fuel, crew, maintenance) than manned aircraft. sUAVs can fly much closer to the ground than the manned aircrafts. This project's research will permit rapid and computationally efficient detection of anomalous areas of plastic pollution density and distribution in visual spectrum images using mobile processors. The dynamic path planning algorithms enable automatic mid-flight modification of drone flight plans based on analysis of incoming data from drone-based sensors. Together these algorithms enable autonomous capture of high-value information and immediate delivery of actionable insights from drone missions. This technology will be a significant improvement or complement on existing monitoring technologies or practices, which may involve high labor and equipment costs.

Tags

Voting

14 votes
Ideate
Idea No. 62