The UK experienced its first fatal accident involving tramway passengers in over 55 years in November 2016 when an early morning service on London’s Tramlink came off the tracks, resulting in the deaths of seven passengers and injuries to 62 more. The Rail Accident Investigation Branch (RAIB) concluded in December 2017 that the likely cause was a ‘micro sleep’ as the driver of the 05.55 service from New Addington arrived at the sharp Sandilands curve at more than three times the posted speed limit, the tram leaving the rails and overturning.
As a result, the RAIB made 15 recommendations with two related directly to excess speed and the responsibilities of the driver. Recommendation 3 asks the industry to look closely at high-risk areas such as the transitions between high- and slower-speed running, requesting operators, owners and infrastructure managers to “review, develop, and provide a programme for installing suitable measures to automatically reduce tram speeds if they approach higher risk locations at speeds which could result in derailment or overturning.”
Recommendation 4 calls for an investigation into solutions that can swiftly and reliably detect driver inattention and take appropriate action when such situations occur: “Such responses might include an alarm to alert the tram driver and/or the application of the tram brakes. The research and evaluation should include considering use of in-cab CCTV to facilitate the investigation of incidents.”
With technology to address both recommendations already readily available, research has been undertaken by technology and human factors specialist Ian Rowe Associates on behalf of UKTram and the new Light Rail Safety Standards Board into how these systems work, their benefits and drawbacks and how they could be applied to diminish the chances of a similar event happening again.
This initial research established two areas for examination: AVSM (Automatic Vehicle Speed Monitoring) and Driver Inattention systems. There are many products in each of these categories, with some addressing both.
There are two main types of AVSM solution: GPS/GNSS ‘digital’ systems and those involving infrastructure modifications and additional equipment such as trackside balises. Both have their advantages and disadvantages and are worthy of exploration here.
As many modern trams and LRVs already feature sophisticated On-Train Monitoring Recorders (OTMR) that log large amounts of data, such as speed and direction of travel, this can be used as the basis for safety enhancements such as those laid out in
RAIB recommendation 3.
GPS/GNSS solutions have greater flexibility in terms of adding or amending speed limitations quickly and easily (for example with temporary or emergency restrictions),
as well as other driver advisory functions such as Correct Side Door Enabling (CSDE). They also require minimal supplementary onboard equipment, thus lowering cost and speeding up installation and maintenance. The major weakness is that location accuracy of GPS/GNSS in isolation is around ten metres, but when this is augmented using inputs from the tram’s onboard systems it can be improved to within one metre.
Another consideration is that such systems generate huge amounts of data that needs to be stored securely, managed and analysed, requiring more sophisticated back office hardware and greater processing power. The upside is that this wealth of information can be used for a range of other valuable purposes, such as driver efficiency monitoring and training, timetable efficiency and ride quality improvements, predictive vehicle and maintenance management, automatic door opening and closure and traction power consumption.
Balise-based solutions offer more established, and higher SIL-rated technology. More common to heavy rail, they are generally pared-back signalling systems and therefore give a platform for further automation (if deemed necessary, or desirable). The hardware is low maintenance, robust and ‘fail safe’, although this comes at a higher capital cost as it necessitates significant investment in vehicle and trackside equipment and, in some instances, its own independent power supply.
Both offer driver alerts and reporting to the system’s control centre, or designated remote staff, and can be further linked to a vehicle’s braking systems to reduce speed in the event of overspeed events. This could be applying the brakes in a progressive manner alongside a warning until the tram is below the speed restriction for modest infractions, or to bring the vehicle to a complete halt if there is a more serious incident.
Other scrutiny needs to be given to how any incident data is to be transmitted securely – for example using the vehicle’s onboard Wi-Fi, or perhaps using an additional GSM element. All these facets need to be factored in when assessing cost and suitability.
Operating any public transport vehicle can be a stressful occupation, with many potential distractions in environments that encompass busy streets and varied interactions with other highway users, be they pedestrians, cyclists or drivers of other vehicles.
Fatigue is one cause of reduced attentiveness that can be particularly hazardous and although there are many causes of fatigue, the primary one is a lack of sleep. There are also proven links between a lack of sleep and high blood pressure, obesity, diabetes and depression and workers who sleep less than 7-8 hours in any given 24-hour period are deemed to be at a higher risk of fatigue-related incidents. Sleep disorders affect more and more people each year, and many, such as sleep apnoea and restless leg syndrome, go undiagnosed. These can play a major part in drowsiness and fatigue.
As cities move ever closer to 24-hour public transport services, there is a greater danger of disrupted sleep patterns for drivers and other operational staff and a build-up of ‘sleep debt’ that can lead to tiredness-induced errors of judgement. As a consequence of shift rotation patterns, the early hours of the morning are seen to be the most vulnerable time of the day, closely followed by the mid-afternoon.
Fatigue is not the only cause of inattention of course and there are other states of degraded performance to consider such as impairment from alcohol or drugs etc; these are all very serious issues, but the implementation of vigilance/inattention devices may lead to early detection of such states.
As well as technological solutions, many of the risks associated with fatigue can be controlled through a proper Fatigue Risk Management System (FRMS) in the same manner as other health and safety issues.
An effective FRMS will benefit both the company and individual workers and should form part of the process when implementing any of the systems described here.
Nonetheless, as city environments become more complex and demanding on the individual, systems that monitor the alertness and attentiveness of drivers of all types of road vehicles have come to the forefront. These can be ‘native’ devices and equipment developed specifically for transport, but also using technology imported from other industries and areas of modern life.
Common factors for analysis when examining implementation include:
• How is driver inattention detected?
• How is the driver alerted?
• If no remedial action is taken, can the system intervene and apply a vehicle’s brakes if necessary?
• Does the functionality exist to notify a local control centre or other designated person in real-time?
• Does the system record data?
One simplified solution is the emergence of ‘wearable’ tech such as wristbands and ear, eye and headgear. Although increasingly part of everyday life for consumers, their commercial use is limited – but this is growing.
While such personal devices have the obvious advantages of being relatively inexpensive to purchase and simple to configure, they also offer a number of downsides. Eyewear needs to be tailored to fit the individual user, particularly if they require prescription spectacles. Earbuds can be uncomfortable when worn for long periods, as well as being impractical for those with hearing aids. Hats and headgear also offer potential impracticalities, especially in drivers’ cabs which are predominantly glass and get hot on summer days.
Some such apparatus needs a consistent link to a mobile device, which could prove another distraction, and may need separate batteries which means control and management of charge. They could also be easily misplaced or forgotten.
The two other main approaches to monitoring inattention and fatigue are therefore either Percentage of Eyelid Closure – known as PERCLOS – or facial analysis systems and Task Monitoring (e.g. Driver Safety Device (DSD) or Traction Brake Controller (TBC) movement) equipment. With these, when inattention is detected audible and visual alerts are given and/or vibration of the drivers’ seat. As with AVSM, some can also be configured to apply the brakes if no action is taken after the initial warning.
PERCLOS and facial analysis use infrared cameras to detect signs of fatigue such as pupil size, blink rate and frequency and head position. They are generally simple to install and allow camera output to be stored and reviewed in incident investigations. The disadvantages are occurrences of false positive detection, for example in busy city centres where the driver has to turn their head more to look for potential hazards, and the extra potential costs associated with interfacing with vehicle systems to enable brake application. There are also question marks over whether some systems are as effective for drivers who wear sunglasses on bright days, and extra time may be needed to calibrate some of the systems on the market to individual drivers.
The acceptance of such solutions by drivers should also be considered – for example, how will the data be stored and used? Is the system compatible with the work environment? Is it an active or a passive monitoring system? What is required therefore is an unobtrusive, early warning system that tracks the early stages of fatigue or distraction; to achieve this, it needs to deliver information in real-time and be easy to customise and configure for individual network and driver needs.
Task monitoring solutions rely on the driver performing a dynamic task within a defined timescale to establish consciousness – for example, employing controls such as the TBC, indicators or bell. This differs from a ‘Dead Man’s Switch’ where it may be possible for the driver to lose consciousness while maintaining pressure on the switch.
Downsides may be that the addition of the drivers’ workload forms another distraction, resulting in false positives, and there is a risk of repetitive required tasks becoming a motor response. Unlike PERCLOS and facial analysis systems, there is no provision to review camera footage for use in incident investigations without creating a hybrid system. However, as a more unobtrusive addition to the cab, task monitoring has a higher probability of driver acceptance.
The single biggest cause of transportation-related accidents is human error. While it is obvious that no driver goes onto shift looking to be involved in an accident, the main ‘errors’ found in a 2016 study conducted by the US Department of Transportation were a failure to operate equipment according to safety protocols and mandated regulations, improper instructions from other crew members and distraction/inattention.
So while full automation of tramway and light rail operations may be some years away (see panel, right), it is important to remember that all of the technology mentioned is designed to support and assist operators and reduce risk, without removing overall control or safety responsibility. Therefore it is vital to consider a number of human factors in the implementation of any new safety-related technology that would affect either the cab or control centre: these include physical ergonomics, usability, conflicts with other equipment/alarms and task load.
Managing the acceptance, or resistance to change, is a principal challenge, but if an appropriate solution is chosen and implemented well – with proven and easily demonstrable benefits – this should become an easier ‘sell’ to those managing day-to-day services. Once accepted, it is still important to manage expectations and clarify responsibilities to ensure that drivers do not come to rely on the system and that procedures are put in place for if it fails. This is where the importance of robust training and education regimes cannot be understated.
Technology in this area is developing at a rapid pace due to an increased interest in vehicle and systems automation globally – mainly from the automotive sector – so it is important that any solution under review is considered with this in mind. With new solutions coming to market all the time,
it is important to view the whole picture.
Added value is also something that needs to be closely factored in. Focusing purely on safety may lead to one set of conclusions but, thinking more holistically, many of the systems described here can provide valuable additional insight into both vehicle and driver performance and behaviour. Such insights can lead to improved efficiency and inform planning, design and procurement decisions, as well as enhancing safety.
The final conclusion is that there is no ‘one size fits all’. Every network has its own characteristics, so the ability to tailor the solution is essential. When evaluating the options it is important to consider the following:
• Infrastructure and vehicle modification requirements
• Practicalities of installation – time, complexity and cost
• Manufacturer install or retrofit
• Data capture and usage
• Further intervention – what, when and how?
It is imperative to therefore remember than people are still the most important element in any transportation system, so any proposed improvement must be assessed in accordance with the common socio-technical model (see diagram, above). Only then will we see the maximum benefits.
UKTram has recently undertaken an initial review of a number of systems that address both RAIB recommendations 3 and 4. The trade body for British Isles tramway and light rail systems is taking this research to the laboratory trials stage later this year. Find out more about UKTram’s work at www.uktram.com
Grateful thanks are due to Ian Rowe of IRAL and James Hammett of UKTram for their input into this article.
Article appeared originally in TAUT 978 (June 2019).