Bus only lane clogged with cars
Image: SounderBruce

Automatic cameras have been approved by the state legislature to monitor speeds, discourage drivers from running red lights and penalize cars not yielding to stopped school buses. But so far, they are not authorized to keep cars out of transit-only lanes by consistently ticketing offending vehicles. (A previous STB post explored the path to using automatic cameras in Seattle.)

According to Mark Hallenbeck, director of the Washington State Transportation Center, using cameras to enforce transit-only lanes is a much more difficult undertaking than it might seem and complicated also by privacy concerns. Here’s what Hallenbeck had to say, transcribed and edited for clarity and length.

LG: What are some hurdles to using automatic cameras to ticket vehicles using transit-only lanes?

MH: Okay, so first the problem comes from the geographic problem of interweaving lanes. For example, look at Highway 99, which uses what’s called Business Access Transit (BAT) lanes to solve the problem of how to make right turns across the bus lane. You don’t want people to have to drive three blocks out of the way in order to come back around. The reality is, there’s no good way to make a right turn across the (transit-only) lane and be safe. So the solution is, you let cars in the (transit-only) lane, so they can make right turns into the shopping centers. Then, of course, you want people — as they come out of those shopping centers — to make a right turn onto the street.

Once you allow that, now you’ve given permission for cars to be in that lane to make the right turns. The problem is now, is how far are you allowed to really drive in the (BAT) lane and can I (the camera) watch you long enough to actually determine if you are in fact a violator? You can’t just say car bad, car fine. Determining what is a violation makes the technology much more difficult.

Unlike with school buses (with onboard cameras), where the technological opportunity to use a camera to catch (offending drivers) is really simple. The camera points in a particular place, and it’s only on when the (bus) stop sign switch is on, and then it’s triggered by something going by.

If you wanted to use the same kind of technology for the BAT lane, you need video versus photo.

LG: Which then can lead to privacy concerns?

MH: Agencies are highly reluctant to just capture video and this has been true from day one. Storing and dealing with video is brutally expensive, as the Seattle Police Department will tell you about body cameras. So the transit agencies kind of look at that, and say ‘well… let’s wait a little bit.’ If you store it, it’s accessible via the Freedom of Information Act, which means you have to be able to not just store it, but you have to be able to give it to people.

It’s not just grabbing files and stuffing them on the hard drive, though even that would be expensive: video takes a lot of disk space. But it’s also how do I retrieve the information and how complicated is that? And then, by the way, how else will other people use the video, if in fact they can get it through the Freedom of Information Act?

LG: Let’s say the privacy issues using video were somehow solved: could cameras be used to enforce HOV lanes?

MH: For the most part, video and still images can’t see inside a car. (Earlier in the interview, Hallenbeck said the angle of the windshield, degree of window tinting and location of the sun all influences the position a human eye or cameras needs to be located to see inside a moving car.) Human observers trying to look inside a car use their eyes to follow the car until the angle is just right so they can see in. But using video or still images tends to be a disaster for occupancy counting, because the camera is fixed at a particular location which means at any given time, the camera position is good for some vehicles and not others, which is not a way to do enforcement.

You can’t use infrared (IR) either, because IR doesn’t see through glass. So the technology they’ve been playing with is near-infrared. Near-infrared can go through glass and gets away from many of the problems that visible light has associated with glare. How accurate? It’s in the high 80s, low 90s (percent) if carefully calibrated. Which is okay for indicating whether a policeman down the way should pull over a car to physically look inside, but it’s not the same thing as writing an automated ticket.

LG: So we’re pretty far from using cameras to catch distracted drivers and enforce the state’s new distracted driving law.

MH: Not going to happen. Enforcement of that law is not what’s going to make it happen. What will make distracted driving go away is the same thing that made drunk driving really drop, finally. And that is individual peer pressure and whether society accepts it, and right now society accepts it.  

LG: What I find really interesting about cameras is the ability to reduce bias in ticketing.

MH: No, it just changes the bias. The bias becomes, well can I see in these vehicles but not those vehicles? That bias is quite likely different than if I pull over all the black people and I don’t pull over the white people. But you’ll see well inside everybody’s cars who didn’t tint their windows, and you won’t see very well inside the cars of  people who did tint their windows. So those technological problems will create biases in the data set that you have.

LG: Do you think manual enforcement is better to educate drivers, rather than getting a ticket in the mail?

MH: I’m not sure getting a letter in the mail has nearly the same impact as having an officer standing there. But I think the fine structure has a far bigger impact. So maybe the first fine is a warning, the second one is a fine, and the third one is a really big fine. I think that has much more of an impact. To me, that’s a higher level of deterrent than whether there’s a cop or a picture. The bigger deterrent is, what’s the likelihood of you being caught and what’s the level of social deterrence, whether that’s the size of the fine or whether that’s how people view you.

52 Replies to “Challenges to Automatic Enforcement: Q&A with Mark Hallenbeck”

  1. We have a lot of BAT lanes, but we also have bus-only lanes, where a car is not allowed, ever. Can we at least automatically enforce the law on those roads via the camera, or is the problem more about BAT lane violators?

    1. Bam. This. Just because it doesn’t work on BAT lanes, which are honestly not where the main problem is, doesn’t mean it can’t work on 100% transit areas.

    2. This would totally work great on Spring and Battery streets. Oh and Montlake Blvd (past Pacific Pl).

      Also. Can’t we also use this on 3rd Ave during rush hours? You know, if the cameras capture the same car driving down the street for over three blocks…that’d be pretty obviously a violation.

      1. Also. Can’t we also use this on 3rd Ave during rush hours? You know, if the cameras capture the same car driving down the street for over three blocks…that’d be pretty obviously a violation.

        I think the problem is the three blocks. It is the BAT lane situation all over again.

  2. Center lanes for buses. If the road’s wide enough, center platforms. If we don’t want left side doors, make bus lanes counterflow. With very effective barriers.

    But since center platforms leave passengers exposed to traffic and weather, put zones at far-side curbs, with signals holding car traffic behind the intersection ’till bus crosses diagonally to the curb. Same when bus departs.

    Signals- preferably triggered by the bus driver-sense bus approaching the zone, so diagonal crossing is clear when arrive. Might work best if buses entered a reserved stretch in platoons, same as planned if not implemented, for DSTT.

    Both requiring and allowing fewer and longer stations. Demanding separate local service running either General Purpose or their own curb lanes. Whether budget likes it or not. Same for the amount of driver training, traffic familiarizing, and expert signal control, maybe needing human controllers.

    Well, yeah, I guess that is a Busway. At least we’ll have something it’s not false advertising to call Bus Rapid Transit. But while we’re working on it, can we just have signals clear right turn lane as bus approaches, and hold ’til bus clears.

    All above is for major arterials. For Downtown Seattle when the 41 and the 550 get daylighted, give all private automobiles full priority in parking stalls at transit centers and in their owners’ garages and driveways.

    Nobody loses. Because unlike rush hour traffic right now, let alone when above routes join the fun, at least everybody in the CBD will at least be moving at walking speed.

    Mark Dublin

    Mark Dublin

    1. Well if we’re going to do all that, might as well put down some track and call it light rail no? Passengers have a smoother ride and you can string multiple cars together

      1. It really depends on the particulars. There are several advantages to bus based improvements over a light rail system:

        1) It can be done for less money. A light rail system has to do all of what Mark described, and then some. You need to add rail and much bigger stops. Just adding bigger stops can be expensive.

        2) You can’t run it up steep hills. That right there kills a lot of ideas for this city.

        3) You can’t just build the improvements where they are needed — you need to add rail everywhere. It is very common to have a route which has only a small section that is terrible or a series of routes that converge onto a common, congested area. Fix that area, and all those lines would be fine. But with rail, you would have to run rail to all those routes or ask people to transfer. The former is very expensive (and you get nothing out of most of it) while the latter isn’t as good for the rider (it takes them longer to get where they want to go).

        4) It is less flexible. Changing bus stops is very cheap — moving rail lines isn’t.

        Jarrett Walker talks about some of the trade-offs. For example, Brisbane serves as a good example of why busways make sense for some cities (http://humantransit.org/2009/11/brisbane-bus-rapid-transit-soars.html — the “Update” section is a great summary that answers the exact question you asked). Another great post is this one, where he discusses streetcars (http://humantransit.org/2009/07/streetcars-an-inconvenient-truth.html). There is a lot of stuff in there, and a lot of it is subtle. He takes great pains to explain the caveats — the subtleties — to his argument. Given the fact that subtlety is in short supply, I can understand why.

  3. Another thought: I happened to run across this video (https://youtu.be/4-B4IdVDHvY) which showed cops enforcing similar laws (in this case, “blocking the box”). One thing that is obvious when watching the video is that is that looks like a pain in the butt for officers. They have to go out there, wave over the car, send the car somewhere else, then write the ticket. This looks like a case where a camera would be really helpful. This would be more of a hybrid system. The cop walks out, takes a picture (or two) then goes back to waiting for the next violator. The ticket arrives in the mail. This could be done by meter maids, really. It would be nowhere near as cheap as an automated camera system, but still a lot cheaper than today.

    1. YES. Let’s use automated enforcement at 5th & Pine, where the “NO TURNS” signs there are probably the most ignored signs in Seattle.

      And Broadway & Denny and University Way & 45th for no-left-turn violations.

  4. Why can’t we have a camera manually operated by the bus driver? Have it pointed forward and if the bus driver sees a car violating the rules (i.e. staying in the BAT lane for multiple blocks), they push a button and the camera takes a picture. Seems pretty simple to me.

      1. So a bus driver isn’t supposed to use his horn, either? We are talking about pushing a button, similar to honking. I would imagine one would be followed by the other. In other words, the bus driver honks, waves for the guy to get out of the way, and if the car driver ignores him (or gives the finger) the bus driver presses the button and takes the picture.

      2. Can only bust one car at a time. Refer to the picture. The bus driver cam is probably fixed and could take a pic of the car in front when it should take a pic of two cars in front. Still, it is better than throwing up your hands which is essentially what this article does.

      3. Ross,

        Use a left foot pedal like the old dimmer buttons on cars pre-1980. Or, more in line with transit, a dead man pedal on a locomotive or streetcar

        Oh, sorry. I didn’t mean to use a profanity…..

      4. I don’t think that it has to be a big deal. Just have them push a button then send the last minute of footage back to base and dole out a fine.

    1. This is exactly how it should work. The bus driver pushes a button which flags the video for review by a LEO. The officer then reviews the tape and judges whether or not the law was broken, issuing a ticket if it was.

    2. While we’re at it, an app for bike riders to upload video clips to SPD parking enforcement for Ubers waiting for riders in bike lanes.

  5. Solution to the “have to capture video” problem for BAT lanes:

    A pair of cameras, a block (or more) apart. In order for a citation to be issued, the same vehicle must be captured by both cameras.

    That took me ~5 minutes to think up, and I’m not even trying. C’mon, Hallenbeck, you get paid for this stuff.

    1. Sounds good, but it might not be that easy. I don’t know how the automatic tolling works, for example. It takes a picture, but then what? Does the software automatically get the license plate number, or does a human look at the picture. My guess is it is the latter. Someone spends their day looking at pictures and then typing in the license numbers.

      If that is the case, then you are asking the software to do something new (recognize the same car, a couple blocks apart). It might not be that hard, but I have no idea if someone has done that yet.

      1. Yes, the software “automatically gets the license number”. Software can pick individual faces out of a crowd now, if it’s looking for a particular suspect. How much easier to read a license plate which is a standard sized, reflective rectangle mounted between one and four feet from the ground?

        Your a tech guy, Ross. Keep up, man.

      2. You know that we are more than capable of reading license plates with a camera, right? It’s be commonplace for years.

        I can assure you that the state does not have a team of people transcribing the license plate of every car that goes through the 405 toll lanes or the 520 bridge.

      3. Yeah, I’m a tech guy, but I know that getting that last bit of certainty is difficult. There is a difference between technology that works pretty good and technology you can count on. Voice recognition has made great strides, and is quite remarkable, yet the court system still uses stenographers. Mr. Hallenbeck mentioned another bit of technology, used to determine how many people are in the car. Like voice recognition, is is pretty darn good — “high 80s, low 90s (percent)” — but you can’t send out tickets based on it. You need a manual backup system, and that is what they use.

        As I said, I’m not certain this is done automatically. You would assume so, but manual backup is pretty common. The more I think about it, though, the more I think that you could easilly create a very good manual backup system. Let’s say the software is 90%. So they take two pictures and are ready to issue the ticket. But before this gets sent out, someone double checks, and throws out the mismatches. Problem solved (no one has to fight a ticket they don’t deserve).

        The false negatives don’t matter. You aren’t catching everyone, but you are still catching way more than before.

        Like the red light cameras, everyone who gets a ticket has photographic evidence of the infraction, and can fight it in court. That makes it different than the HOV situation, because they can’t get good pictures of the situation (their 90% rate is based on near-infrared). So it makes more sense to just pull those people over and then double check.

        So, yeah, I think the idea makes sense. It should definitely be explored. My guess is that they aren’t doing it because no one else is doing it. It would probably cost some money to set this up, and no one wants to spend on this. You would probably have trouble getting a grant, because unlike red light cameras, and school zones, this isn’t life and death.

      4. I have an anecdote about license plate readers. The Vancouver Police have a car with a tower on it kind of like a Google car. They drive around retail parking lots all day and it reads license plates. It automatically checks them against the national database of stolen vehicles.

        My car was stolen one night from in front of my house. It’s an Acura Integra which is basically a Civic in drag. Anyway, Honda used to have very low security keys. The ignition can usually be turned by a “jiggle key”. Don’t ask; it’s what the cop called it.

        Anyway, we reported it stolen at about 8:30 in the morning and we got called about 4:30 PM. The reader car had driven by it in a Fred Meyer parking lot and raised the alarm. The cop watched the car and when the kid who stole it got in he walked up and arrested him.

        Pretty cool.

        Also, you do know that GoodToGo uses license plate readers, right? They send tolls based on photos from them. I do think that an police officer has to view the video and OK the bill, but since most people who use the drive through lanes actually have GoToGo accounts, the police don’t have to OK many bills.

        The same would be true of the bus videos. The software would find “candidates” and the cops would OK the ticket. One officer could probably OK all the candidates for a given day in a couple of hours.

      5. And it’s not “no one else”. Every bus in San Francisco carries a front facing camera. I don’t know if they’re on all the time that the bus is in a bus lane or are activated by the driver when someone actively violates the lane.

        I do know that their introduction raised a terrific hullabaloo and the craven Board of Supervisors canceled the program after six months. But during that six months the buses moved markedly faster.

        There is a concerted political movement to remove those who voted “Yes” for removal, so they may be back.

      6. Oh, I just read that the State trumped the City! AB 1287 makes “TOLE” permanent. Wow, what good news.

        Of course, the Washington State legislature is more likely to forbid Seattle for doing it than requiring it.

      7. Also, you do know that GoodToGo uses license plate readers, right? They send tolls based on photos from them. I do think that an police officer has to view the video and OK the bill

        Yep, that’s basically what I said. Someone has to give the final OK, as opposed to being fully automated. I think that would work here, which is why I said as much in my second comment. Maybe I wasn’t clear — I changed my mind — I think Lack Thereof’s idea could work. At worst you have someone double check the info (as you suggest they do with tolls) to make sure you don’t improperly fine someone. The big problem (as I see it) is that we would be first (as far as I know).

        And it’s not “no one else”. Every bus in San Francisco carries a front facing camera. I don’t know if they’re on all the time that the bus is in a bus lane or are activated by the driver when someone actively violates the lane.

        Yeah, but that’s a different system. That is one based on a bus, and even that was controversial (apparently). It is my understanding that the bus driver has to basically target someone, and only when they are parked (as opposed to driving too long in a BAT lane).

        This idea sounds better, and is more elegant. Just have a series of cameras, every block or so. If you pass by more than one, you get a ticket. You really have only a couple things to do:

        1) Make it clear how long you can drive in the BAT lane (before turning). You could mark the cameras clearly, and then publicize the “don’t pass more than one” rule. That would be a lot less of a judgement call than things are now.

        2) Build the system. Again, I think the problem is that we would be first. It sounds very cheap, but you still have to spend the money (on cameras, the software and maybe the person double checking the tickets). We should be able to afford it, but it is much easier politically (and cheaper) if you simply copy someone else. I know, it would be great if we were first (civic pride and whatnot) but generally speaking, we lag the rest of the country when it comes to basically everything transit and transportation related.

        Plus, as much as this issue gets people riled up, it is nothing like the other uses of cameras. Tolling allows people to avoid a toll booth, which is a convenient bonus for drivers. With the HOV lane violations and stolen cars, the cops are notified, as opposed to an automatic ticket. The only really automatic thing that drivers might not like is red light and school zone enforcement. You are penalizing drivers, without the chance to explain their situation at the time of the infraction. But everyone understands the importance of safety in both of those situations, so they embrace it.

        But BAT lane violators? My guess is that automatic ticketing would not be nearly as popular. Some guy works his way over to the BAT lane a bit early, takes a right with no bus in sight, and then gets nailed for a ticket. Now people are pissed at both the automatic ticketing and the BAT lanes. That would be the fear, whether it is justified or not.

        I’m not saying it shouldn’t happen, or couldn’t happen, but the fact that no one else is doing it (or has developed the system) is probably the biggest problem.

  6. Thank you for publishing this. Camera enforcement/fines in the bus lanes is the next logical step in transit reliability, in my opinion.

    Some clarifying questions (I hope) from a layperson:

    1) I see the problem with turn lanes that are shared with BAT lanes. What’s preventing us from starting a pilot in areas that don’t have these sorts of turn lanes, and gather data on how it works + clear out some parts of our network and collect some cash? Other commentators have already suggested a few potential spots above.

    2) What is the reason for looking INSIDE the car? If I understand the mechanism correctly, we only need to take a picture of the license plate and send the bill to the address on the record. Shouldn’t it be handled just like the Good to Go tolls on the bridge?

    3) With respect that this introduces a person who needs to be paid a salary into the picture (and thus might be cost-prohibitive), can you think of any other problem with having buses fitted with cameras which are turned on once entering the downtown core, and having a human filter through the footage within the next calendar week or so, preserving for permanent record only the parts of the video which show a clear violation, deleting the remainder, and sending out a ticket to the address on record? If having every Metro bus so be-camera’d is a cost problem, I would start with the RR buses.

    4) Not a question so much as a proposal – I think the first places to put cameras/turn on cameras should be where buses enter the Mercer Mess. This could be sold as helping out SOV drivers as well, as the focus for punitive tickets could be for drivers who are “blocking the box” and thereby selfishly making everyone else’s commute longer. Getting political buy-in for cameras here could be the springboard towards expanding them throughout the transit network.

    Any thoughts/opinions welcome…

    1. Re: #2, he is talking specifically about HOV lanes here. You need to see inside the car to see if there is more than one person (and thus an HOV lane violation).

    2. As Frank says, for (2) he was clearly talking about HOV enforcement. That said, generally speaking for a traffic infraction to be , there has to be evidence that the ticketed person actually committed the violation.

      State law generally provides that tolls are payable by the owner of the car. For traffic safety camera tickets, there is a special provision in state law (RCW 46.63.075) that allows a refutable presumption that the car’s registered owner is operating it. I don’t see how either of these would apply here without a change to state law.

  7. On BAT lanes, does anybody know if business can have their truck unloading or doing something in the bat lane (or a ups blocking the lane)? Because this happens a lot on aurora.

  8. Mr Hallenbeck.

    Mount forward facing cameras on the buses and activate them only when a similar forward facing low-power lidar (it’s cheap now because RoboCars) says there’s something in the way of the bus

    As to the privacy, WHO THE HELL CARES if the selfish, unthinking bastards are outed? If you’re driving in a BAT lane farther than a block or in a reserved Red lane any time, you ought to be on the six o’clock news in a shaming segment. Think of it as mobile stocks.

    1. With your Facebook profile photo as an inset to the “evidence” photo……

      [P.S. That “your” really should be the generic “one”, but woo-hoo that would get jeers on this website!]

    2. And if the jerk violator’s spouse’s divorce lawyer gets some ju-ju from a FOIA request, well, the guy shouldn’t have been driving in the BAT lane with his paramour.

      1. https://en.wikipedia.org/wiki/Juju

        Richard, an attorney will need an Département d’Agriculture Haïtien endorsement to have the goat mailed to him. Before he even thinks of going to the clinic.

        Problem is that since Haitian law also requires the juju practitioner to give your wife’s lawyer ALL the information, he won’t get a single barbecued goat rib back when he finds out the violator’s paramour is his own wife.

        So better just stay with Jim Garner at the Rockford agency.

        However BAT reference definitely offers us both revenue and violation deterrence. http://bacardicuba.net/the-history-of-bacardis-iconic-bat-logo/

        Especially if this campaign catches on, Bacardi will fund BRT lanes nationwide. Great label for the front of a bus! Which also has advantage of having an already guilty conscience see a giant red and gold bat in his rear view mirror.

        And best of all, kids will have something to yell at their parents to let them ride on besides LINK.


  9. How about this: camera directed car license plate level at the Bus Only words in the lane. Camera takes a pic every minute of so. Most minutes will have nothing or the vehicle is moving too fast. Some minutes will have a clear car license with the Bus Only sign, and you nab those. You aren’t going to capture every infraction, just the car that lost the gamble and is stopped in traffic. The infuriating part is the stopped part or worse, using the lane as a pickup passenger lane.

    1. A bus with 50 or 100 people onboard doesnt create traffic, its every one of those people in a single occupant vehicle that creates traffic.

      1. 50-100 cars could never get through in a light cycle or two anyway. And of course the reason a bus does on occasion block a box is because of SOV car congestion.

      2. It’s not just the cars that would get through the intersection that the bus is blocking, but its also the cars (and busses) in other intersections that can’t proceed because the first light is blocked. Multiple blocks are affected by one intersection.

      3. @William C. – It happens nearly every day, frequently, in the afternoons at 6th and Stewart. The “block” to the west between 6th and Westlake is quite short and the one between Westlake and 5th is shorter, and both cars and buses frequently block the intersection. When buses do it it’s not uncommon to have 6th blocked completely until the signals change and Stewart can move.

        It does not matter much exactly what is blocking them to buses (and cars, trucks, etc.) that are stuck upstream in cross traffic – traffic is stuck regardless.

        Stewart is a perfect example of why it would be nice to have left-hand doors/left side BAT lanes on some of the wider one-way streets; after the stop between 7th and 8th several routes are trying to do the weave from the BAT lane on the right all the way over to the left to make their southbound turns; when there is any sort of traffic at all this becomes very difficult there. As a lot of those routes are suburban, though, left-hand doors take up valuable seating so the weave-and-block will continue. :(

    2. Buses block the box because drivers are inconsiderate. Usually if the the bus doesn’t attempt to proceed through the intersection and tries to wait for enough space, drivers turning right on red will fill the space and the bus goes no where. Banning right on red would do much to alleviate this.

  10. Is nobody concerned about the proliferation of cameras in public spaces and the corrosion of privacy? Once there’s a pervasive surveillance infrastructure in place don’t you think it’ll be abused?

  11. They have police standing at Minor and Howell at weekday PM rush hour and they do NOTHING about cars driving in the Howell bus lane and its always the douchebag in the oversized pickup truck that doesnt give a f* about anyone else.

  12. What is unsaid:

    1. Getting tickets for driving in the bus lane would have pushback because the bus lanes are already unpopular with certain people. Red light cameras are also unpopular because well… people don’t like getting tickets.
    2. It’s easier and cheaper to do nothing than to do something. Probably this just isn’t a high priority, especially at the state level.

    As others have mentioned, while they couldn’t cover BAT lanes, they could cover bus only lanes. Also, we could start transitioning some of our BAT lanes to bus lanes. Some need to be BAT, but not all. Sometimes it is ok to prevent cars from turning on a particular street.

    In other words, with this like many issues political pressure can make a difference.

  13. “Once you allow that, now you’ve given permission for cars to be in that lane to make the right turns. The problem is now, is how far are you allowed to really drive in the (BAT) lane and can I (the camera) watch you long enough to actually determine if you are in fact a violator? You can’t just say car bad, car fine. Determining what is a violation makes the technology much more difficult.”

    Just say you can’t cross a signed perpendicular road while in a BAT lane. Cameras at the block, if you proceed through then you get a ticket.

Comments are closed.