Join us to participate in the upcoming 2019 City:One Challenge. 

Safe Cars - Safe Streets

A pilot to test a driver alert system and develop safety standards for vehicles to provide safer streets for pedestrians and cyclists.

Photo of Jennifer Sayer
2 1

Written by

The road to zero traffic injuries and fatalities

Only 10% of Pittsburghers walk and 2 - 4 % cycle on their daily commute. Still, they were involved in 40% of fatal collisions in 2015 (BikePHG, 2016). According to the PennDOT, an accident is rarely caused by dangerous behaviors of pedestrians and cyclists, but rather a result of aggressive driving or inattentiveness of the driver. Investment in improving pedestrian and cycling infrastructure and increasing vehicle safety standards will have a large effect on improving road safety for all and will bring Pittsburgh closer to zero traffic fatalities. Pedestrian and cyclist friendly driving attitudes also encourages more people to walk and cycle in Pittsburgh in the long run.

The slogans of many self-driving car developers promise zero road incidents. We think that this should not only be a promise but a requirement made by cities and their citizens if we are to replace the human driver with artificial intelligence. In fact, we can start developing intelligent safety measures already today, before reaching full autonomy.

Humanising autonomy today - an intelligent driver alert system


We are developing a pedestrian intent prediction platform to increase the safety of vulnerable road users (i.e., pedestrians, cyclists, people with disabilities, the elderly, and children) in dense, complex urban environments. With common dashboard cameras, we can provide the driver with intelligent warnings about cyclists and pedestrians in their path. Our software combines deep learning and behavioral science to analyze video footage, recognize people, and predict their intents. Are they about to cross the street?; Did they see the bus?; Is this person distracted, perhaps on the phone?.

Adopting our pedestrian intent prediction platform in Buses and HGVs could increase safety for pedestrians and cyclists while avoiding driver cognitive overload. This platform will accomplish the work that more cameras, mirrors, and sensors would do, but in a more accurate, seamless and efficient way without overwhelming or further distracting the driver. If anything, our technology will ensure that the driver can remain alert and attentive on the road, while benefiting from accurate and advanced detection technology at blind spots.

We want to make intelligent safety standards for vehicles an integral part of future mobility. While road space allocation is renegotiated and at the same time becoming more convoluted with new mobility options, the vehicle’s awareness of more vulnerable traffic participants should be enhanced - from “people watching out for vehicles” to “vehicles watching out for people”.

In the scope of the Ford City of Tomorrow Challenge, we can integrate our software and cameras in buses running through central Pittsburgh. The project will bring insight into near-term and long-term measures to make vehicles safer and understand behaviors of people around the vehicle (when do people feel comfortable passing?; Why did that person not see the bus and what can we do to change that?; How does the driver’s driving style change?). This information will be valuable for urban policy development in the face of changing vehicle technology. Testing and developing a vehicle safety standard and driver alert system can help the Pittsburgh Department of Mobility and Infrastructure (DOMI) achieve their goal of zero deaths and serious injuries on the city’s streets.

Describe who will use your solution (1,000 characters)

In the first instance, during the pilot phase, only the Port Authority, who runs the BRT services in Pittsburgh. They would implement video cameras and our software on their buses. But finally, Pittsburgh’s citizens will be users and beneficiaries of this solution, namely cyclists, pedestrians, other vehicles, and the bus drivers. After verifying the technology during the pilot, the new safety standard can be adopted in buses and HGVs as well as inform safety regulation and policy development for autonomous vehicles.

Describe your solution's stage of development

  • Prototype - you have built a prototype and tested it with potential users

Tell us about your team or organization (500 characters)

https://www.humanisingautonomy.com/ We are a team of deep learning engineers, behavioral data scientists, designers, software developers, and urban planners; that makes 9 in total and we are growing.

Size of your team or organization

  • 2-10

Funding Request

  • $100,000

Describe how you would pilot your idea (1000 characters)

Use of funding Installation of cameras on buses: The specifications have to be further developed in collaboration with the Port Authority, but this could be the range: Two buses each on two different routes with 4-6 cameras on each bus. Initial data collection to train AI model: Collect data from buses via an online platform or directly from the bus and train our model with the collected data. Integration with existing or new alert system: Installation in the four buses. Final report: Analysis of findings, learnings, and proposals for further development of safety standard and policy Labour and expenses The pilot depends on collaborations with public and private parties; the Port Authority, the DOMI, and possibly a further entity who provides hardware for the alert system. The product that is developed during the pilot, can be adopted widely by HGV, bus, and other vehicle manufacturers. The report will support city officials in transport and urban policy development

Describe how you would measure the success of your pilot (1000 characters)

Key performance indicators are: - Percentage of successful pedestrian and cyclist recognition and intent prediction - Consistency among vehicles and routes - Feedback from drivers - Number of situations where the alert averted an incident - CBA of the driver alert system

2 comments

Join the conversation:

Comment
Photo of DeletedUser
Team

DeletedUser

An idea to put into vehicles is if it is in motion all cellphone use will be disabled. When stopped a call may be made. Thus humans wouldn't be able to call until motion remain at 0. This would increase social interaction in say an autotmated taxi/Uber.
One other point I addressed and solved in my design is speed and the miss use of it. should you want check out my post: Tired of tickets, towing and crashing? Enable the future!!!

View all comments