Cameras in Perth's Taxis

by Dr. Ian Radbone
Transport Systems Centre
University of South Australia
19 February 1998



I visited Perth from Wednesday, 11 February to Saturday, 14 February to look at that city's experience with the introduction of compulsory cameras in taxis, as well as examine other transport-related matters. Information about cameras in taxis was gained from interviews with the following people:

Conversations with 12 other taxi drivers were also partly designed to find out their views about cameras in cabs. I brought back a copy of a training video demonstrating the use of the cameras.

The following report discusses the thinking that went into the decision to adopt the Raywood system. It then discusses the system's characteristics and capabilities before providing evidence of its success.

Process of decision-making

Two serious injuries to taxi drivers in 1995 created such a level of concern that the government decided to sponsor a one day taxi safety summit in February 1996. The level of concern is indicated by the fact that there were 200 attendees. An appreciation of the situation was created by a number of speakers in the morning. The afternoon was devoted to workshops on what to do about it.

Four broad strategies were considered:

  1. the use of a purpose built taxi,
  2. the installation of screens,
  3. the introduction of a global positioning system (GPS) and
  4. the installation of cameras in taxis.

While the installation of a camera was not necessarily considered the most effective option, it was broadly supported because of its immediate feasibility and non-intrusiveness. The Minister agreed to support this approach by making the cameras compulsory and subsidizing their installation.

A purpose-built taxi is obviously only feasible in the longer term. I was surprised at the widespread support for a London-style cab as a much more effective strategy to both protect as well as deter. This may be been due to a recent marketing push in Perth by a company promoting these. The marketing effort will be renewed at an expo of vehicle related items in Perth in September. (The company's product can be viewed on the web:

Screens and GPS were abruptly ruled out. The WA Worksafe was and still is a force for screens. The Taxi Industry Board (TIB) is keeping a data base of incidents in an effort to demonstrate that cameras have improved things partly in an effort to allay a push for screens. According to one informant about 40 screens had been installed in Perth taxis in the past but none were installed at the moment. Other arguments about screens concern the expense of the testing process to meet ADR 69, the limited range of vehicles that currently approved screens can be fitted to and the fact that screens make it more difficult to hear the passengers.

I have the impression that GPS was vetoed by the management of the dominant radio company Swan Taxis. One interesting comment I heard was that the police did not want GPS, for fear of encouraging a vigilante element within the taxi industry.

Perth is the first city in the world to have compulsory cameras. I was told that there are some cameras installed in Brisbane on a voluntary basis. It was argued that if cameras were not compulsory those without cameras would chosen as targets for crime.

A panel of ten was established to determine appropriate specifications, including one consumer representative and a specialist engineer. This panel was soon being approached by a number of potential suppliers. An initial 32 was reduced to about ten when the decision was made to only consider systems which were digital (solid state). When combined with a budgetary considerations this ruled out video. There were two or three videos in the initial tests. I was told that there were high rates of failure and maintenance involved. Another argument against videos was that the temptation to keep recordings of sexual behaviour in the back seat would be a threat to privacy though of course there would have been procedures to prevent this.

None of the products being offered in the first tender met all the specifications but four vendors were given the opportunity to spend some time developing a product which did. As a result of this second round of tenders the Raywood system was chosen (See it at, apparently on grounds of price and reliability. There followed a period of extensive testing and fine tuning, especially of the housing of the computer. Three cars were destroyed by fire as part of the testing. (It is feared that a serious offender would try to destroy the evidence by burning the car.)

The cost of the camera system installed was $1570. The government agreed that $1000 of this should be paid out of its Taxi Industry Development Fund. (The Fund is made up of remaining financial assets when the old taxi board was replaced by the Department of Transport, plus proceeds from the tendering of new taxi licences.) The rest has been paid for by the licence owner. It was pointed out that the system is an asset which enhances the value of the car. It can be transferred to a new vehicle. Being solid state it does not wear out, though of course it may become obsolete. And of course the owner's expense is tax deductible.

The installation of the cameras began in October 1997 and all but a handful of cabs had them installed by mid December, the date by which they were to be compulsory.

Operations of cameras

The system consists of a camera mounted above the rear vision mirror connected to a 24 megabyte computer attached to the car. There are two infra-red devices allowing clear images at night; one around the camera itself and one next to the interior light. The computer can be located in a number of positions, such as under the wheel arches. The box that contains it is tamper proof and water proof. It hangs from the car by nylon screws which would melt if the car was set on fire. A series of tests found that by dropping to the ground the insulated box could protect the computer from the heat of a car on fire.

Solid state digital technology has been chosen because solid state is virtually maintenance free, can more easily withstand vibrations and jolts and because digital images can be manipulated (e.g. blown up easily) to help identification.

The cameras themselves (including the infrared) cost $385 a relatively small part of the overall cost. In some cases vans have a second camera mounted mid way along the vehicle's roof, though this is not compulsory. The power to store the image needs to be independent of the car. This is supplied by AA batteries which will need to be changed every two years.

The Raywood system takes a series of snapshots, both automatically and driver activated. Images can be taken through the following actions:

129 digital images can be stored. Subsequent images replace previous ones once the 129 limit is reached, so normally images will be held only the past four trips or so. Drivers can freeze the previous 30 images by holding down a button. They can unfreeze these just as easily.

The industry itself chose to have the images taken and held in this way. Changing the software can change the deployment.

In the case of an alarm, with all images frozen, the computer has to be reset. This can be done only at a depot of a radio company or taxi management company, or at Raywood. The fee for this can vary but is about $5. Once hooked up it takes only a few seconds. The alarm is activated by pressing a button on a switch that sticks out from the steering column. The driver can also "freeze" a series of photos which would not be over-ridden unless he or she unfreezes them.

As is well known in the taxi industry, the usual position of the alarm button on the floor where is can be pressed by the driver's foot creates many false alarms. (Although ironically for short drivers usually women it has proved difficult to reach.) Always a nuisance, a false alarm is more serious when it is connected with the camera system and renders the camera unable to take further images. The new arrangements have eliminated accidental triggering of the alarm and deterred frivolous alarms. Perhaps it has gone too far. Of course it is always possible to restore a foot alarm which could notify the base without activating the camera. One driver I spoke to suggested two alarm buttons were needed as now drivers were too inhibited to press the alarm button.

Perth's alarm systems requires the alarm button to be pressed for five seconds. While this prevents accidental alarms, the attacker would probably be able to see this being done. (The five seconds could be easily reduced by reprograming.) While chatting with drivers on a night rank I was told by one driver that he had had a gun pointed at him a week or so previously but (I think) had not used the alarm. I'm inclined to disbelieve the story, but the portrayal of panic was clear enough to make me wonder about whether the action required to initiate the alarm is always feasible.

Downloading of images takes over half an hour. (Midway charges $40 to do this including a $20 refundable deposit for the zip disk which holds the downloaded images.) Once again it can only be done at a limited number of centres. The zip disk is then sent in a sealed envelope to the police. The downloading centre can see only the last image on its computer screen. Only the police and Raywood have the technology to see all the images.

The cost and bother means that in practice the system will be unlikely to be used to catch fare evaders. It has been argued that drivers need to be deterred from frivolous downloads. The protocols require that only the police can see the images. Making it costly to download is a way of preventing police time being wasted.

The one basic requirement of the system when it was set up was that it be able to obtain one clear image of the passenger(s) to identify an alleged offender. It is not expected that the camera will record an actual crime. For example in the case of one of the bashings that prompted the summit the victim was dragged out of the car before being bashed.

The camera is effective at taking the one clear image required. The infra-red arrangements do an excellent job, even at night. The photo following was taken in darkness, demonstrating the infra-red qualities. It is of an actual break-in. The photo resulted in a prosecution.

The system seems to be robust which it should be given the testing process. Ten to fifteen cameras have failed and have had to be replaced so far. It is non-intrusive, yet obvious to the passengers.

Passenger security

The way it is set up means that the driver can always over-ride images, simply by continually opening and closing the door. So it is not designed for passenger security. It is surprising that more attention was not given to passenger security given the damage caused to revenue by fears of passenger security. The WA Department of Transport estimates the taxi industry lost 30% of its business ($60 million) in 1996 and $45 million in 1997 due to a series of murders of young women in which the finger of suspicion has been pointed at Perth's taxi drivers. The fact that the camera is on board gives a sense of security to many passengers, even if it is false.

It is probably possible to store images in a way so that they could not be over-ridden by the driver. For example with file compression, a single photo of passengers from the last 70 or so trips could be held if they were triggered only by the action of turning on the meter. More could be held if image compression was used or lower "density" images recorded, but both options imply a possible loss of picture quality. However a driver could over-ride even these images by turning the meter on and off continually. Some thought would need to be given to find a way of having images taken which was independent of driver action. So far not much thought has been devoted to the issue because it has not been a priority.

Future enhancements

The system has been described as "GPS friendly" and "mobile friendly". The former means that the image can include the GPS location if GPS was used. Currently no radio company has GPS in Perth. Images could be sent to base, though this would probably mean a second radio band. It would also need someone at the base to determine what action (if any) should be taken in the event of a possible alarm. But if images were sent to base merely to record them for passenger security purposes this would be no problem. Images could be captured by third parties unless other methods were adopted to counteract this.

The verdict

The cameras have been compulsory for two months. What's the evidence of effectiveness so far? The TIB data base has recorded a drop in reported incidents but the numbers are too small to be statistically significant at this stage. It is also too early to see any convictions stemming from the evidence of the cameras. I understand about fifteen downloads have been sent to the police, or which about ten are being used in investigations. Of course the real value is the deterrence effect. As the Raywood representative noted, the best value for money improvement to the system he would recommend would be to make bigger signs warning passengers they are being photographed.

Senior people in the industry that I spoke to have been uniformly happy with the decision and the product. The Taxi Industry Board has not recorded a single complaint, either from industry or passengers. Resentment among owners was dampened by having $1000 of the $1570 cost paid for out of a government-controlled taxi fund.

Drivers also appear to be happy. Obviously two and a half days in Perth did not allow me to undertake any real survey of drivers attitudes, but during that time I spoke to 16 drivers, only one of whom (a day shift owner-driver) thought the cameras to be a waste of money. He argued that driver behaviour had not changed, that cameras were only useful for serious offences and that the risk of these was so low that it did not justify the expense. In his view the best defence against the more common minor matters was better training and selection of drivers.

Another day shift owner-driver said he had not pushed the alarm button in 15 years, but would prefer screens to really protect the driver. But he knew he was in a minority. He admitted to an incident a couple of days previously when a couple of "low life" hopped in the cab, noted the absence of a camera and said "good, now we can do what we like", but soon quieted down when he claimed to have a new type of hidden camera.

Three drivers I spoke to about peak period taxis said that they didn't want them originally but now think they are a good way to keep the passengers behaved. . But it should be pointed out that they would be expected to support cameras, as the idea of their licence is to carry multiple passengers late at night. One noted that the camera can soon pay for itself in the costs of cleaning up the cab, as long as the driver can bluff the passengers that the images can easily be used for prosecution. Another pointed out that the camera prevents women threatening to allege sexual advances by the driver as a means of avoiding paying the fare.

The fact that the images are not accompanied by sound did not seem to bother anyone. Pressing the alarm button enables the radio company operator to listen. Nor was it a concern that the images were not video. It was thought that photos taken every second would be sufficient for prosecution.

Clearly the theme of deterrence is important. So is the perception of both drivers and passengers. One driver felt the main benefit of the cameras to be the safety they provided for passengers. We know this to be false, but while the drivers believe it to be true it will have some effect.


Clearly the adoption of the same system in Adelaide would not be a bad decision, at least in terms of popularity within the industry. Whether it would be most effective depends on careful comparisons with the alternatives.

If the decision was made to install cameras we would need to recognize the on-going costs of using the cameras and develop the infrastructure and protocols for downloading, resetting, and maintenance. Fortunately the Perth experience could help us here.

Back to the Taxi-L Safety Page