Technical Overview

 

Local Street Maintenance

 ARGO's experimental Computer Vision techniques at work to automatically identify street defects.

ARGO's experimental Computer Vision techniques at work to automatically identify street defects.

 ARGO's 2016 survey of Syracuse, NY's street conditions. Over 110,000 images and upto 500 miles of street condition data were collected using 1 vehicle in just 10 days!

ARGO's 2016 survey of Syracuse, NY's street conditions. Over 110,000 images and upto 500 miles of street condition data were collected using 1 vehicle in just 10 days!

Street maintenance is the most visible example of local government performance.

While CA cities and counties face a $78 billion shortfall over the next decade to adequately maintain the existing network of local streets and roads, approaches to understanding the conditions of our street networks remain a relic of the past.

Today, data collection for resource allocation and performance evaluation exists at two unsustainable extremes. At one end of the spectrum, cities rely on the low-touch, windshield surveys. These approaches provide little ground truth and attempts at network-wide collection are inevitably constrained by human subjectivity and fatigue. At the other end, cities have turned to high-touch, laser and lidar vans. But this ground truth comes with a price tag. It can take up to 3 years to cover a street network of a large city like Los Angeles once.

In recognition of the current situation, California Senate Bill No. 1 has committed $15 billion dollars for local street and road maintenance conditional on performance tracking. To use these public dollars respectably, the future of street maintenance must embrace a cost-effective approach to frequent, comprehensive surveys with sufficient ground truthing. The StDC calls this new approach Street QUality IDentification (SQUID).

A SQUID survey is as simple as a ride through the city. Accelerometer readings and street imagery are automatically collected to generate a network-wide map of ride quality, preserving ground truth at low cost. In addition, computer vision, the same set of proven techniques that automatically identify faces on Facebook and let Teslas autonomously stay in their lane, can identify potholes and other street defects even prior to inspection of imagery by city agencies to help focus our attention.

By using SQUID to reduce the time between surveys and their costs by an order of magnitude, the StDC is shifting us from a reactive, whack-a-pothole blitzkrieg to a paradigm of truly anticipatory street maintenance.

Further Reading

Our 2016 efforts profiled in national media to advocate for a digital survey to ground truth local street maintenance.


Real Time Transit Data

 A $10 Raspberry Pi Zero is a fully-functional wifi-equipped computer capable of retrieving its location.

A $10 Raspberry Pi Zero is a fully-functional wifi-equipped computer capable of retrieving its location.

 A recent ride on a New York City bus using the $10 Raspberry Pi device and Wifi-positioning to retrieve the bus's location. A citywide system can be built to develop real time bus schedule infrastructure at the fraction of what it costs today.

A recent ride on a New York City bus using the $10 Raspberry Pi device and Wifi-positioning to retrieve the bus's location. A citywide system can be built to develop real time bus schedule infrastructure at the fraction of what it costs today.

Access to real-time transit information has been linked to overall satisfaction with transit serviceincreases in ridership, and substantial increases in farebox revenue. If cities could simply increase practical availability to transit information, they could achieve outcomes similar to increases in transit service itself. Encouragingly, this missing layer of coordination between providers and users amounts to a conceptually simple piece of technology.

Less encouragingly, the legacy technology in this space can be exorbitantly expensive. NYC’s bus-tracking GPS system has been quoted at $20k per bus. After a protracted 3 year deployment process, Melbourne, a moderately-sized city, was forced to suspend rollout of it’s original bus-tracking system in 2013 at only 30% coverage due to unreasonable operating costs. While all cities and citizens could benefit from a real-time transit information system in principle, this is not an option for all cities in practice. As forward-looking municipalities, we have to find a new approach.

To address the situation facing our cities, the StDC is leveraging the recent explosion in low-cost sensor technology to develop a BusTime4All. Simple microcomputers with internet access enable wifi-positioning at the accuracy of a few meters at costs orders of magnitude cheaper than legacy GPS systems. With SB1 committing $7.5 billion for transit operations and capital, and the StDC’s legacy-breaking approach, CA cities have a unique opportunity to increase accessibility to their transit systems as well as increase farebox revenue from this increased ridership.

Further Reading


Preparation for Autonomous Vehicles

 ARGO's computer vision techniques can be used to detect and identify the quality of pavement markings in a city. This can help cities better allocate maintenance resources and get ready for Autonomous vehicles

ARGO's computer vision techniques can be used to detect and identify the quality of pavement markings in a city. This can help cities better allocate maintenance resources and get ready for Autonomous vehicles

 Keeping pace with advances in self-driving engineering, we are building capabilities to help cities reimagine service delivery for Autonomous futures. This GIF shows a lane-detection example via ARGOnaut,  Graham Henke .

Keeping pace with advances in self-driving engineering, we are building capabilities to help cities reimagine service delivery for Autonomous futures.
This GIF shows a lane-detection example via ARGOnaut, Graham Henke.

In the next decade, vehicles capable of full autonomy outside of only the most extreme conditions will hit the consumer market. In the next two decades, these vehicles are predicted to outnumber non-autonomous vehicles. Among the myriad changes these transitions will bring, there is the potential to reduce inefficient land use and emissions as well as unleash massive demand for automobile travel among adult non-drivers, the elderly, and the disabled. Adoption by these groups could increase overall vehicle travel in the United States by as much as 14 percent. There is a massive opportunity here for cities to serve their citizens and for citizens to become more economically productive within their cities.  See the excellent Bloomberg report on "taming the autonomous vehicle" for more information. 

But this future is not preordained. Volvo’s CEO, Lex Kerssemakers, exclaimed to LA Mayor Eric Garcetti that he needed to  “...paint the bloody roads…” when his semi-autonomous prototype failed to drive. 

Tesla’s CEO Elon Musk has called the issue of faded lane markings “crazy". Cities have an active role to play in drawing out the benefits of autonomous vehicle adoption. StDC will bring its network of technologists and mobility experts to bear on realizing the potential future that is available to us.

The technology that powers our SQUID framework for local street maintenance has been extended to detect fading lane markings. To assess preparedness for vehicle autonomy will require a street network-wide assessment of lane marking quality. We have a window of opportunity for cities to proactively prepare for autonomous vehicles so this historic technology serves the public interest.

Further Reading