In a prior NITC project that ended in early 2019,we successfully built and deployed a hardware and software ‘product’, called ‘Bike Connect’, that allowed people on bike to give hands-free advance information to an upcoming traffic signal based on speed and direction of travel to increase the likelihood the signal would be green or be quicker to get to green upon arrival. In this new project, we propose scaling up this work in three ways: 1) deploying additional hardware to two additional variable timed signals to create a bicycle - and scooter - instigated green wave; 2) integrate fixed-time signal changing data into real time information for these micromobility users to give them an audio countdown tool for any signal, as well as some signal status info as they are moving to create their own green waves and 3) deploy, verify, and explain the accuracy of (or not) machine learning applications that can be applied to entire municipal-level traffic control systems anywhere in the country.
This work will use Eugene as an applied laboratory, focusing on two key corridors in particular: 1) 13th Street where a new 2-way, $2.5 million, 1.5 mile cycle track and signal upgrade project connecting downtown Eugene with the University of Oregon campus will be completed in Fall, 2019; and 2) the Alder Street greenway, a 1+ mile corridor that includes the initial Bike Connect enabled signal along a 2-way buffered bikeway serving as a key link between a residential neighborhood, the University of Oregon, and other regional destinations along a regional bicycle network. These two corridors are the most heavily utilized by bikes in Eugene, itself one of the nation’s top cycling cities, and will likely be among the most popular corridors for scooter use when that system is deployed in late summer / early fall 2019.
Our expected methods and results:
1. Bike Connect deployment: we will deploy the system to two additional signals along the Alder corridor, measuring ‘time to green’ for app users vs. non-users and the ability to create an entire ‘green wave’ under different conditions (e.g. riding solo vs. en masse);
2. Fixed-time signal communication to cyclists and scooter users: we will create new, low-cost, bike-mountable hardware (or an app that mimics such eventual hardware) that informs people on bike and scooter of traffic signal changes to aid in timing a green light or helping inform users at a stopped light of the time remaining before green. We anticipate being able to measure user behavior in two main ways: speed adjustment to minimize complete stops and changes in behavior crossing streets against a red light;
3. Machine learning: We will apply machine learning to traffic signal data to help predict phasing of actuated signals so that micromobility users can have real time information about the likelihood of green lights at upcoming signals, in addition to providing real time data on fixed-timed signals.
In the end, we expect three outcomes of value to researchers and practitioners nationally:
1. Viability of a vehicle to infrastructure communication system that influences signal behavior when the vehicle is a bicycle or scooter;
2. Viability of infrastructure to bicycle communication that is scalable and works within existing signal systems of cities; and
3. Insight into and daylighting of the application of machine learning to street-scaled, traffic signal systems for the benefit of users on bike or scooter.