Bicycle Adaptive Traffic Signal Timing

I’m only an occasional bicycle rider but I have always thought that it would be nice if traffic signals could respond appropriately to the special characteristics of bicycles. Bicycles are different from cars. Among other differences are bicycles do not have air bags nor 5 mph bumpers. Forgetting that though, there are other differences between bicycles and motorized vehicles that need to be addressed. The three most important differences are acceleration, speed and stopping distance. A motorized vehicle can do all three better.

Still, there are good reasons to ride a bicycle. Better fuel economy, good exercise, less pollution, etc. are three very good reasons for riding a bicycle, especially for shorter expeditions.

Many people opt to utilize their bicycles and current rules of the road allow a mix of bicycles and motorized vehicles on the same roads. For many years there were no special considerations given to bicycles. More recently, some signals have installed special bicycle push buttons to provide bicycle timing. This requires the button to be located adjacent to the bicycle path or for the bicyclist to go to the pushbutton sometimes requiring dismounting. Not altogether convenient.

Many agencies made their inductive loop detectors sensitive to bicycles. The MUTCD has even been adjusted identifying bicycle detection requirements. This allows the bicycle to be detected but the timing provided is usually set for motorized vehicles rather than bicycles. There are often also other problems encountered unless special loops are utilized. A loop designed to pickup bicycles often will not detect trucks or SUV and or may detect cars in adjacent lanes causing improper traffic signal timing.

Often, two detector systems are installed in the traffic lane, one to detect bicycles one to detect other vehicles. Obviously this can be a costly method of detection. There are two things that would make things better for bicycles without adversely affecting motorized vehicles. One is bicycle adaptive timing. Adjust the traffic signal green and/or clearance time to be appropriate for the slower bicycle. In order to do this, a good means of detecting bicycles needs to be utilized. It would be better if the same loop detector system could discriminate a bicycle from other vehicles either by the signature of the actuation, amount of actuation, or special loop location.

Obviously, the easiest way is providing a special lane for bicycles that cars don’t travel over. This can be as simple as installing loops in the bicycle lanes. This does require that there is room for the bicycle lane, that bicycles use it and that cars don’t travel over it such as when making right turns.

A better approach is loop detectors that discriminate bicycle actuations from other vehicles. This is possible by providing a different output when the frequency variation of the loop array is low and a separate output or output condition when the actuation is greater. A typical bicycle causes 10-16hz frequency variation while a small car would be over 20 hz variation. Unfortunately, some other considerations need to be made to prevent false bicycle actuations due to cars passing near the loop rather than over it.

A better method would be to utilize an actuation signal signature difference to identify and discriminate bicycles from other vehicles and then provide the individual outputs for each. This should just be an adaptation of the typical counting detector operation. These detectors utilize loop arrays and only provide a single pulse for each vehicle that leaves the loop array. They have accuracies of greater than 98% and should be capable of providing bicycle discrimination with appropriate programming.

OK, now assume we have appropriate detection. Now, we need to adjust the traffic signal timing for the bicycle while still providing the least impact to the other traffic. There are three considerations. A bicycle starts out and accelerates much more slowly than other vehicles, doesn’t go as fast, and doesn’t stop as quickly. A traffic signal typically provides a minimum green time, a green extension time and a clearance time appropriate for the motorized vehicles. If the traffic controller could provide a different minimum green, green extension and clearance time adjusted appropriately for the lower acceleration and speed of the bicycle. Existing formulas can identify appropriate intervals, the key is having a controller that is capable of timing the intervals.

For example, a typical vehicle minimum green might be 6 seconds, green extension might be 3-6 seconds and a clearance interval 4.5 seconds for a roadway with a speed of 35 mph. When adjusted for a bicycle, the minimum green might be 8-10 seconds, green extension 6-10 seconds and the clearance interval of 5-8 seconds.

An important note here is that the vehicle clearance interval consists of both the yellow interval and the all red interval. The yellow interval should not be extended for the bicycle. The all red interval should provide the additional clearance. This is because motorized vehicle traffic could have difficulty responding to varying yellow clearance intervals since most drivers consider it green time and proceed to travel through the intersection on the yellow.

The Naztec model 920 traffic controller provides such timing. Reno A&E evidently has a bicycle discriminating loop detector system available. I haven’t seen the detector yet, but I’m told it exists and is in use.

Obviously, if the discriminating loop detector works, it really opens up adaptive bicycle timing. Even if a controller didn’t have any special timing capabilities, the detector could be set to provide different extension intervals for the two outputs allowing at least a bicycle appropriate green extension interval.

The real benefit of having a controller with bicycle timing capabilities is the ability to provide the extended clearance time for bicycles. At larger intersections, left turn travel distance can reach up to 180 feet. The amount of time it takes a bicycle to travel that distance could be more than 15 seconds. Typically, the motorized vehicle travels the distance in less than 6 seconds. Obviously, good practice doesn’t allow always having the longer interval for bicycles when they are not present.

To sumarize bicycles and other vehicles are different. Different timing needs exist for each. Methods now exist to provide for those differences.

For more information, try these links:

This is the first draft of this document and will probably be updated as errors and ommissions are identified. If you identify errors or can provide additional information please contact the editor using the contact form on the right sidebar or register and post a comment directly to this article.

Newsletter Request Form

If you are interested in receiving our electronic newsletter, please use the contact form below and include your Name, Email address, and enter your company or organization name and any other comments in the comment area.


The following Vendors help sponsor this organization.

If you’re interested in advertising here, contact us using our contact form. Your add appears both here and in our mailed newsletter. $65 / 6 issues. Ad space is limited.

March 21st, 2007 Meeting

Time: 09:30 a.m.

Location: The meeting will be held at the County Traffic Operation Center, 1505 Schallenberger Road.

Speaker: Ananth Prasad with the County of Santa Clara, Traffic and Electrical Operations will preent information on bicycle adaptive signal timing.

Map Page

 If you clicked on the map button associated with a blog entry, the map below will be focused on the location associated with the entry. If you clicked on the Navigation link for the Map Page then a wide view map is displayed showing many map pins. By clicking a map pin, information about the entry and a link to it will be shown. If there are multiple entries for the same location then only the most recent will be shown.

Bicycle Detection Study Articles

These are links to some older articles related to bicycle detection.

If the links are not active, open the article by selecting the link below:

More PLC Applications In Traffic Signal Operations

Written by Tony Rucker, City of Campbell, CA.

What’s a PLC?

Generally speaking, a Programmable Logic Controller, or PLC, is a solid state control system which has a user-programmable memory for storage of instructions to implement specific tasks, such as: I/O control logic, timing, counting, arithmetic, and data manipulation. PLC’s are generally programmed in what is known as Ladder Logic. This method of programming was established because it could be closely related to hardwired relay logic that PLC’s were developed to replace. PLC’s are used in many industries for process control such as parts manufacturing, lumber mills, food processing and auto manufacturing.

The City of Campbell uses PLC’s to supplement the operational needs of several signalized intersections and were chosen because of their high performance-to-cost ratio over relays and 24 volt external logic cards. The PLC receives inputs from the signal controller’s NEMA outputs, makes logic decisions based on its operator-written stored program, and then outputs commands to the signal controller’s NEMA inputs. The PLC used in Campbell for traffic signal operation, the “IDEC Micro-1® “, is a fixed, 8 input/6 output, “brick type” PLC that can be expanded to a total of 16 inputs and 12 outputs. Although several I/O types are available, the model used in Campbell has “source” inputs and “sink” outputs, so that like a NEMA signal controller, it recognizes a ground as a “true” input and outputs a ground as a “true” output, in reference to the cabinet’s +24 volt dc power supply. It has EEPROM memory capable of storing 600 steps of user program and numerous internal logic components such as “AND” gates, “OR” gates, latches, 80 timers, and 48 counters. It can be programmed with either a hand-held “Boolean type” loader with LCD display or ladder logic software that runs on an IBM, or equivalent computer. Other PLC models and brands are available that can be modularly expanded as needed to provide up to 512 I/O points, floating point math, high speed counting, line voltage I/O and analog I/O. These more costly, higher-end units can be configured with RS-232/422/485 serial interfaces for peer-to-peer networking and telephone modem interfacing, providing remote control and monitoring. Of course, the concept of using external logic cards has been available for some time from controller manufacturers, e.g. “Econologic”, “MultiLogic”, etc.. However, this PLC only occupies 1/4 cubic foot of space and for only $300 it can replace relays and timers costing 10 times as much and taking up 50 times more space.

A Sample PLC Application in Campbell

Refer to Figure 1 below. The intersection of Hamilton Avenue and Eden Avenue utilizes 6 phases. Hamilton, the arterial, utilizes phases 2 for EB Thru, 1 for WB LT, 5 for EB LT, 6 for WB. Eden, the SB side street, is phase 3 and a commercial driveway is phase 4 for NB. Pedestrian movements are with phases 2 for EB, 6 for WB and 3 for SB. Figure 1 shows the signal phase sequence and intersection configuration. The goal was to provide alternate pedestrian timing for the SB movement phase 3 which could be enabled by a crossing guard. The guard would flip a toggle switch inside the locked police door to enable or disable the alternate ped timing. The alternate ped timing was necessary for the guard to be able to accompany a large group of students across the wide arterial and still be able to return to the appointed post before the arrival of the next large group. The operation was designed with the following logic: Utilize phase 3 ped timing during “normal” ped timing operation, allowing the 3 ped to operate concurrently as it normally does with phase 3 vehicle. However, when “alternate” ped timing is selected by the crossing guard, phase 8 ped timing will be selected to time AND display the “WALK” and “DON’T WALK” display for the SB ped movement, concurrently with phase 3 for the vehicles. The logic will also insure that this “alternate” phase 8 ped will only serve with phase 3 SB and NOT with phase 4 NB, as that combination (8 ped with 4 vehicle) would be allowed in normal NEMA controller operation and configuration. At this intersection, however, NB and SB are each protected movements with left turn green arrows, hence, the SB ped can operate concurrently ONLY with the SB vehicle, phase 3.

Figure 1.

The following is a description of the ladder logic program in the PLC (refer to figure 2):

Line 0 allows that the enabling of the alternate ped occurs only when the toggle switch (Input #1) is closed AND phase 3 is NOT in green (Input #7). Thus, the alternate ped is enabled when internal relay #401 is SET.

Line 3 allows that the disabling of the alternate ped occurs only when the toggle switch (Input #1) is NOT closed and phase 3 is NOT green (Input #7). Thus, the alternate ped is disabled when internal relay #401 is RESET.

Line 6 allows that the walk output (Output #200) to the loadswitch will be driven by phase 3 walk (Input #2) when the alternate ped is disabled (relay #401 has NOT been SET) OR by phase 8 walk (Input #4) when the alternate ped is enabled (relay #401 has been SET).

Similarly, line 12 allows that the don’t walk output (Output #201) to the loadswitch will be driven by phase 3 don’t walk (Input #3) when the alternate ped is disabled (relay #401 has NOT been SET) OR by phase 8 don’t walk (Input #5) when the alternate ped is enabled (relay #401 has been SET).

Line 18 allows that phase 8 will be omitted (Output #202) when NOT in phase 3 green (Input #7) OR when the alternate ped is disabled (relay #401 has NOT been SET).

Line 21 allows that a hold will be placed on phase 3 (Output #203) when phase 8 is displaying walk (Input #4) OR ped clearance (Input #6). To clarify, ped clearance is TRUE whenever a NEMA controller is timing the flashing don’t walk interval.

Figure 2.

The cabinet wiring was modified and the Conflict Monitor was programmed to additionally protect the SB “WALK” from NB phase 4. Also, a jumper plug that mates with the PLC’s connector was added so that if the PLC must be removed, the technician connects the cabinet harness to the jumper plug instead of to the PLC. This passes the NEMA controller’s phase 3 “WALK” and “DON’T WALK” outputs directly to the loadswitch so that the phase 3 ped operates “normally” (concurrently) with phase 3 vehicle until the PLC is returned to service.

Another Distinct Advantage

Another distinct advantage of PLC’s is that they can be easily modified or reprogrammed to meet changing intersection operational needs without having to purchase and install more connectors, sockets, cards, or relays. Usually the required modifications are limited to running a couple of wires between I/O points of the NEMA controller and the PLC. After the new logic operation has been checked thoroughly in the shop for the correct operation, the technician can download the revised program in the field with a loader or a laptop PC and then field check the operation to insure its conformance. However, if the PLC-to-Cabinet wiring is installed as described below, in a matter of seconds you can replace the existing PLC with a spare PLC already pre-programmed with the revision.

PLC Wiring in Controller Cabinet

PLC’s are usually hardwired in their more familiar process control environment such as plants and factories. This is generally not acceptable in the realm of traffic signals, as technicians prefer the modular, connectorized concept of the controller, conflict monitor and other cabinet equipment applied to all active components. This facilitates maintenance and decreases down time. To achieve this, the City of Campbell selected a 24 pin “AMP” brand circular plastic connector (CPC) cable system. The chassis-mounted plug with male contacts is mounted on a small piece of sheet aluminum that is mounted with stand-offs to the PLC’s mounting holes. Standard nylon-jacketed, 19 strand #22 AWG wire is used to connect the pins from the rear of the CPC to the terminal points on the PLC. Only the AC power, neutral and ground should be carried on shielded #18 AWG cable. A mating CPC receptacle with female contacts is wired with conductors of sufficient length to be connected to the appropriate controller I/O terminals in the cabinet. A 35 mm DIN rail is secured to the cabinet’s inside wall where appropriate and then the PLC is snapped in place. Terminal blocks, fuse holders and relay bases are also available that are specifically designed to snap onto the DIN rail, further facilitating the process. Installation is completed by screwing on the female harness to the PLC. A standard wire list of PLC I/O function assignments was adopted early and adhered to, so that PLC’s are interchangeable throughout the City. A PLC can be removed and replaced with another unit in less than 30 seconds. The only unique feature of the PLC installation at a site is the PLC’s internal logic program.


This article describes the use of reasonably priced, readily available alternative control tools to enhance the signal operation beyond what a typical NEMA signal controller offers. A PLC application has been successful in Campbell but it should be noted that, to successfully program and install a PLC requires above-average knowledge of NEMA controller operational specifications in addition to PLC programming. An experienced signal technician can obtain the knowledge required to implement PLC operation by taking courses in PLC operation and Ladder Logic Programming, beginning with basics and fundamentals and then graduating to the more advanced classes. Most major manufacturers offer factory courses and many are free. Contact the manufacturer’s representative for the PLC of your choice for details.

It would be ideal if a signal controller had a user programmable area for a traffic engineer or signal technician to program the desired logical operation without having to rely on external devices. But until that time arrives, you might consider using a PLC to enhance or optimize your signal operation. There are four PLC’s currently in use in Campbell. They have provided continuous, trouble-free operation since installation in 1994. And although there have been numerous power outages in that time frame, not once has any of the PLC’s caused a problem.

170 Watchdog Timer

Written by Steve Claypool, Caltrans

Bottom line: if you are purchasing a 170 system, or new or replacement 210 conflict monitors for the 170 system, specify a 1.5 second watchdog timer. Caltrans began specifying 1.5 second monitors in 1989, New York State DOT began specifying them in 1991. Some of the newest monitors are available with switch selectable 1 or 1.5 second watchdog monitors, and many older monitors can be modified to 1.5 seconds. This does not affect the green conflict timing or 24 volt power loss timing, both of which remain at .5 seconds.

The problem with the 1 second w/d timers is that, in some cases, they cause recurring “false” calls, putting an intersection into w/d flash when there is no hardware or software malfunction. Ninety to ninety five percent of 170 systems rarely, if ever, experience this problem, even with the 1 second w/d monitor. The “problem” intersections will go on w/d flash intermittently, sometimes with predictable regularity, from as little as about once a year, to as much as once or more per week. We know that sometimes the problem is caused by utility failures; a few problem intersections have been cured when a loose or corroded neutral was found and repaired. We suspect other sources of utility caused w/d flashes, but are not sure of the exact nature of the failure. Caltrans labs tried to simulate various line distortions, but could not duplicate the w/d failure. We also know that the problem is sometimes in the cabinet wiring. Several instances of recurring false w/d problems have been cured by replacing the output file or the entire cabinet. It would be logical to assume that some recurring problems are caused by a combination of the two causes.

In several cases where output file or cabinet replacement solved a recurring problem, the components were inspected and no apparent malfunction was found. In most cases where line problems were suspected, no identifiable problem was found. Various line filters have been tried over the years with no clear benefit. Several coincidences have been noted; one is that an intersection may develop a recurring “false” w/d problem when there is medium to heavy construction in the immediate area. This may be a factor of the power tools and equipment causing line disturbances, but many times the w/d problem occurs in the late evening or early morning, indicating it may be somehow related to the nature of the temporary service supplying the site. Another coincidence is that a rash of false w/d calls often occur after a widespread power failure. This is particularly bothersome because it often occurs during stormy weather when most signal techs are already busier than normal. *Note: large scale power outages often cause “real” watchdog failures also, either because of bad batteries in the 170 or because of temporary or permanent damage to the processor caused by power surges.

The watchdog monitor monitors a pulse put out by the 170. The “pulse” is written into the program, and changes state every 100 milliseconds. This pulse is output through one of the NPN transistors on the I/O board of the 170. A watchdog fault occurs if there is greater than 1.1 seconds between signal changes (greater than 1.6 seconds in a CALTRANS type monitor). Less than +4 VDC is a LO, and greater than 12 VDC or open circuit is a HI. The extra half second apparently gives the system a little more time to recover from whatever transient problem is causing the false w/d flash.

In the East Bay Area, once we began putting the 1.5 second monitors in the problem intersections, false w/d calls dropped dramatically, from two or three per month to about two or three per year: almost invariably in intersections which still have 1 second monitors. (Even though the standard has changed, 1.5 second monitors are only being introduced in new cabinets or as replacements for defective monitors, so the 1.5 second monitors represent only about 10% of all monitors and were more or less randomly distributed around the area. Once we understood the advantage of the newer monitors, we began to move them to the problem intersections, accounting for the reduced trouble calls.