One of the most effective ways to speed up public transit operations in mixed-flow traffic is transit signal priority (TSP). TSP, in a broad sense, aims to reduce delay arising from waiting for red lights at intersections. This is important because intersections can be a significant cause of public transit slowness. Reducing this delay can help give transit the speed advantage it needs to compete with alternative modes of mobility, especially driving.
The most basic way to implement TSP is through “passive” techniques. In general, passive techniques simply seek to improve traffic and reduce signal-induced delay for all vehicles along a transit route. For example, in Los Angeles, passive TSP is implemented along the Metro E Line’s street running segment along Flower St. in Downtown LA. Here, the signal timings are adjusted such that all traffic on Flower St, including Metro Rail trains, have a green light for a greater long-run proportion of time. Another common passive TSP tactic is traffic signal coordination that helps create a “green wave” along a certain corridor when travelling at a prescribed speed.
Passive TSP can be useful. For example, when the 10 freeway in LA was closed in mid-December, LADOT implemented better signal timings along the aforementioned Flower St corridor in LA. This passive TSP implementation reduced rail travel times along that segment by up to 10%.
However, the really effective implementation is active TSP. Active TSP relies on actually detecting transit vehicles as they approach an intersection. When an intersection’s TSP-enabled traffic signal controller knows a transit vehicle is coming, it changes the traffic signal timing to allow the transit vehicle to proceed through the intersection more quickly. This typically means extending a green light along the transit vehicle’s path if it would otherwise have to wait through a red light, shortening a red light in the transit vehicle’s path, or inserting traffic signal phases that serve the transit vehicle’s path into the light cycle.
It’s important to note that active TSP is different than traffic signal preemption. Traffic signal preemption means a transit vehicle would immediately turn a traffic light in front of it green while stopping all other directions of travel. Traffic signal preemption for transit is not very common, because it can significantly disrupt traffic besides the preempting vehicle. However, preemption implementations exist; a well-known example is TriMet’s MAX light rail in Portland, Oregon. MAX uses signal preemption in downtown Portland because city blocks there are extremely short (literally the length of a 2-car MAX train consist), and without preemption, trains would need to wait at many signals. Traffic signal preemption is more commonly used to allow emergency vehicles to force green lights to appear when their lights and sirens are activated.
Portland seems to be quite forward-thinking in terms of transit signal optimizations. I saw a particularly sophisticated and effective active TSP implementation when I was there a few months ago. TriMet recently rolled out their FX2-Division BRT service. FX2 features these very cool dedicated bus signals, along with active TSP that predicts (!) when a FX2 bus will approach a given intersection and inserts a green bus signal phase into the signal cycle, such that the bus will probably have a green when it actually approaches the intersection. According to TriMet, these dedicated signals plus their active TSP have reduced bus travel times along the BRT corridor by 20+%.
It turns out that Portland uses the same system for active TSP that I’m writing about in this post. It’s called LYT, and I found out about it on my recent tour of the City of San Jose’s traffic management center. This isn’t a sponsored post; I’m just writing about LYT because it’s really cool tech.
As previously mentioned, active TSP only works when traffic signal controllers know that a transit vehicle is approaching. In most cases, this works by mounting some kind of appliance on the signal mast arm and the transit vehicle, such as an infrared gun (on the vehicle) and infrared receiver connected to the signal controller (on the signal mast arm). This works OK, but it’s expensive because it requires new hardware deployment on potentially thousands of vehicles and signals. It’s also limited in effectiveness, because it usually requires line of sight between the vehicle and the traffic signal for vehicle detection technology to work.
LYT takes a totally different approach! It operates based on the assumption that the transit agency already has automatic vehicle location (AVL) equipment on its transit vehicles. This is a reasonable assumption, since many agencies already rely on AVL to provide accurate next-bus/next-train times to customers, and to aid dispatch staff. LYT also assumes that the traffic signals that need active TSP are connected to a networked automated traffic management system (ATMS). The ATMS allows for signal controllers to be reprogrammed remotely. This is also reasonable, since many departments of transportation have recognized that installing an ATMS is far cheaper than sending staff into the field every time a signal needs to be retimed/reprogrammed.
With these two key technologies in play, LYT uses AVL to monitor the location of transit vehicles that leverage TSP. Taking into account data like transit stop locations, historical travel times, traffic congestion, schedules, and vehicle speed, LYT uses AI to predict when a transit vehicle will approach a given intersection. With this information, LYT communicates with the ATMS to tell the signal controller at the intersection to insert a green phase. It aims to give the transit vehicle a perfectly-timed green light when it reaches the intersection. However, this process is intelligent and designed to keep transit perfectly on time, which is possible because LYT has access to transit schedule. That means LYT may elect not to try and give a green light to a transit vehicle that is ahead of schedule. On the other hand, it can try to give green lights more aggressively to transit vehicles that are behind schedule.
The staff at the City of San Jose mentioned that LYT’s technology improved travel times so much that Santa Clara VTA (the main transit agency in San Jose) had to significantly edit their schedules for LYT TSP-enabled routes. I believe ridership also jumped as a result of the improved experience. And here is the crazy part: the LYT rollout for three bus routes in San Jose took only five months!
This is quite possibly one of the coolest yet most practical applications of artificial intelligence I’ve ever seen, let alone in the transportation sector. As someone who uses transit frequently, I know how annoying it is to be constantly stopped at intersections. Given how easy, quick, and cheap it is to roll out LYT, I’d love to see similar technology-driven active TSP be implemented in LA. To me, it seems like this technology is one of those few transportation initiatives that are no-build yet massively impactful.
If you’re curious, you can learn more about LYT here.