Testing of autonomous vehicles has revealed new questions on vehicle safety to be asked
Safety and autonomous vehicles exercised the minds of some of Europe’s senior police officers at the recent 4753 TISPOL European Traffic Police Network Conference in the UK
The European Union looks like missing its target of halving the number of people killed on its roads each year by 2020. Just when European police forces are trying to get back on target, along comes the autonomous vehicle with all its inherent safety issues.
“The target will be missed because there was only a 17% decrease in road fatalities between 2010 and 2015 when [the rate of reduction] should have been at least 30%,” said Rudolf Koronthaly from the EU’s Road Safety Unit.
“We are now in a period of stagnation which started in 2013 and is continuing,” said Koronthaly, who is also a policy officer in the EU’s Directorate-General for Mobility and Transport.
Koronthaly told delegates to the conference that cutbacks in government spending “may well be having an effect” on enforcement capabilities. The EU might retain its “world-leading position” on road safety… but there is no room for complacency, he said.
“There are still far too many fatalities on our roads and the situation for vulnerable road users is not improving anywhere near fast enough,” Koronthaly told the gathering of more than 300 of Europe’s top traffic police officers in Manchester.
TISPOL president chief superintendent Aidan Reid, who is also head of the Garda National Traffic Bureau in the Republic of Ireland, agreed. “We must remain alert and on point if we are to make our targets and we must remain focused on our four key objectives: to be safe, secure, effective and efficient.”
Across the EU during 2015, 26,000 people died in road crashes, according to Koronthaly. A quarter of those incidents took place in urban areas with 20% of the victims below 25 years old. Three-quarters of the victims were men and 45% of the fatalities involved passenger cars. Around 30% involved pedestrians and cyclists and the EU now averages 51 fatalities/million inhabitants.
To hopefully improve things, the EU is bringing forward new laws. “Legislative work underway includes a new professional driver's training directive,” Koronthaly said, “where we are reviewing the training and qualification process for truck and bus drivers.” There is also a new set of tunnel safety and infrastructure safety management directives coming into play and the EU is looking carefully at the issues around vehicle automation and connectivity.
TISPOL president Aidan Reid praised the recent Project Edward - the European Day Without a Road Death - initiative. “I think that success lies in partnership,” he said. “In fact, the whole future of road policing is going to be about partnership.” Multi-disciplinary ideas like Project Edward and how they are promoted and run is clearly the way forward. “We must understand how technology can help us, how we can work with other agencies and how we can better use evidence that has been tested to get the results we need,” he said.
This was a theme echoed in the keynote address by professor Oliver Carsten from the Institute for Transport Studies at the University of Leeds. He questioned how to harness the benefits of new vehicle technology to improve road safety. What sort of legislation should we promote as a result and how should the road safety sector go about lobbying Europe’s Members of Parliament?
Carsten told the conference that there is already an EU General Safety Regulation (GSR) study in place “to consider the potential of crash avoidance technologies to supplement crash mitigation technologies”. It was published in March 2015 and, according to Carsten, it “sets the European regulatory agenda for 2016 onwards”. However, if it is to have any sort of “actual outcome in terms of legislation [it will be] a co-decision of Commission, European Parliament and Council.”
Carsten outlined the “active safety” technologies already in place and, “based on the evidence reviewed, the following measures were considered to be likely to be cost-beneficial and could on that basis be taken into consideration: enhanced AEB with collision mitigation; intelligent speed adaptation; lane keep assist; reversing detection and reversing camera systems; and emergency brake light displays.”
Things are happening fast. “New vehicles sold from 2022 onwards are likely to have Intelligent Speed Assistance as a required fitment,” Carsten told TISPOL. However, “it remains to be seen exactly what type of system the Commission will propose. Hopefully there will be a dramatic increase in speed compliance and less work for the police”.
He remains hopeful that autonomous vehicles could reduce the number of accidents caused by people driving under the influence of alcohol or drugs. “The European Commission estimates that across the EU around 25% of all road deaths are alcohol-related,” he said.
“There were an estimated 240 fatalities in Great Britain as a result of drink-drive crashes in 2013, which equals about 14% of all fatalities [and] in Great Britain during 2014, 0.9% of drivers admitted to driving under the influence of illegal drugs at least once in the previous 12 months.”
Of course, he added, “automated systems do not fall asleep” and “25% of all fatal crashes on motorways and major trunk roads in the UK are sleep-related. In Victoria [Australia], around 20% of all fatal road crashes involve driver fatigue. For Western Australia, it is 30%.”
However, Carsten warned that people “will tend to be more sleepy in highly automated vehicles”. He cannot see autonomous vehicles having any beneficial effects if the systems are based on rapid human intervention in a crisis.
So, for Carsten, one of the biggest questions that the industry needs to answer is “how to manage the handover from automated back to human driving. It is one of the major design challenges,” he said. For instance, “how will the vehicle know that the human is ‘available’ to drive?”
The occupants might not be asleep, but they also might not be paying attention. They could be taking part in “a non-driving task such as video or email. The typical answer is to monitor the human but eye movement cameras are not 100% reliable and don’t work at all for some drivers.”
Carsten urged for a greater understanding of how the autonomous vehicle and its occupants collaborate. “They will need to understand each other’s authority, capabilities and intentions. The human machine interface (HMI) has a crucial role in communicating between the two,” but the perfect system does not yet exist.
Carsten cited a recent fatal crash in the US where the driver of a4071 Tesla was using the car’s autopilot system and hit the side of a truck trailer that was crossing the road. This raises a series of issues.
“The Tesla driver was on autopilot and reportedly driving well over the speed limit. The truck driver did not detect the Tesla or thought he had sufficient time to manoeuvre. The Tesla driver was reportedly watching a movie. The forward vision sensor system on the Tesla reportedly did not detect the truck trailer, which was white against a bright background. The sensor system saw the space under the trailer as empty space (and carried on as if it was driving under a bridge or a gantry).”
As a result the Tesla collided with the side of the trailer at high speed, slicing off the top of the car.
So where might the blame lie in situations such as the Tesla accident? Could it be a sensor failure? Might there be too much vehicle speed? Would the road layout be to blame? Carsten said that the safety lobby is going to need to work on things like: “the speed limit for automated driving; better sensors; requiring side under-ride guards on truck trailers; geofencing … which means preventing the enabling of automated driving on lower-quality roads through use of a digital road map; previewing intersections via a digital map and disabling of automated driving on approach to an intersection; the control of distracting activities where attention is required; and vehicle-to-vehicle communication.”
Carsten said he could also envisage a requirement for an event data recorder, commonly called a black box and similar to ones in airplanes that track the performance of the aircraft.
In the long run, though, autonomous vehicles could be good news. “Driver assistance systems are likely to increase driver rule compliance and automated vehicles will be generally safer,” Carsten concluded “However, there are some real challenges in ensuring that safety is maximised and analysis of electronic data will play a major role in crash investigation.”
Distraction daze
The modern driver is facing an “epidemic of distraction” thanks to mobile phones, tablets, music systems and GPS screens. It is harder for a driver to focus on the road ahead, which is why sergeant Neil Dewson-Smyth of UK’s Cheshire county police launched the ‘don’t stream and drive’ initiative. He is using social medial to encourage drivers to keep their eyes on the road and their hands on the wheel.
“A well-trained driver and a vigilant car is a win-win scenario for the future,” said Neil Greig, director of policy and research at the UK’s Institute of Motorists. But he too warned that a plethora of devices hitting the market is creating a “Wild West in mobility” for users. The institute is also worried about driver awareness versus the car’s level of automation. Who has final responsibility for what? What activities unrelated to driving should be allowed? Where will lie liability and burden of proof in conditional and high automation?
Greig suggested that the car of the future should be benchmarked against the best human drivers. “This is a once-in-a-generation opportunity to ensure that technology replicates the best that humans can be. Let’s aim for 5-star cars driving on 5-star roads controlled by 5-star drivers.”
The more likely outcome, he warned, will be autonomous driving system manufacturers rushing to market with a mish-mash of unproven products. The market could be headed towards a battle similar to that of the VHS versus Betamax struggle for video recording supremacy, he warned. “Consumers will be left in the middle, not knowing who to trust, what to buy and which platform will become the industry standard.”
The European Union looks like missing its target of halving the number of people killed on its roads each year by 2020. Just when European police forces are trying to get back on target, along comes the autonomous vehicle with all its inherent safety issues.
“The target will be missed because there was only a 17% decrease in road fatalities between 2010 and 2015 when [the rate of reduction] should have been at least 30%,” said Rudolf Koronthaly from the EU’s Road Safety Unit.
“We are now in a period of stagnation which started in 2013 and is continuing,” said Koronthaly, who is also a policy officer in the EU’s Directorate-General for Mobility and Transport.
Koronthaly told delegates to the conference that cutbacks in government spending “may well be having an effect” on enforcement capabilities. The EU might retain its “world-leading position” on road safety… but there is no room for complacency, he said.
“There are still far too many fatalities on our roads and the situation for vulnerable road users is not improving anywhere near fast enough,” Koronthaly told the gathering of more than 300 of Europe’s top traffic police officers in Manchester.
TISPOL president chief superintendent Aidan Reid, who is also head of the Garda National Traffic Bureau in the Republic of Ireland, agreed. “We must remain alert and on point if we are to make our targets and we must remain focused on our four key objectives: to be safe, secure, effective and efficient.”
Across the EU during 2015, 26,000 people died in road crashes, according to Koronthaly. A quarter of those incidents took place in urban areas with 20% of the victims below 25 years old. Three-quarters of the victims were men and 45% of the fatalities involved passenger cars. Around 30% involved pedestrians and cyclists and the EU now averages 51 fatalities/million inhabitants.
To hopefully improve things, the EU is bringing forward new laws. “Legislative work underway includes a new professional driver's training directive,” Koronthaly said, “where we are reviewing the training and qualification process for truck and bus drivers.” There is also a new set of tunnel safety and infrastructure safety management directives coming into play and the EU is looking carefully at the issues around vehicle automation and connectivity.
TISPOL president Aidan Reid praised the recent Project Edward - the European Day Without a Road Death - initiative. “I think that success lies in partnership,” he said. “In fact, the whole future of road policing is going to be about partnership.” Multi-disciplinary ideas like Project Edward and how they are promoted and run is clearly the way forward. “We must understand how technology can help us, how we can work with other agencies and how we can better use evidence that has been tested to get the results we need,” he said.
This was a theme echoed in the keynote address by professor Oliver Carsten from the Institute for Transport Studies at the University of Leeds. He questioned how to harness the benefits of new vehicle technology to improve road safety. What sort of legislation should we promote as a result and how should the road safety sector go about lobbying Europe’s Members of Parliament?
Carsten told the conference that there is already an EU General Safety Regulation (GSR) study in place “to consider the potential of crash avoidance technologies to supplement crash mitigation technologies”. It was published in March 2015 and, according to Carsten, it “sets the European regulatory agenda for 2016 onwards”. However, if it is to have any sort of “actual outcome in terms of legislation [it will be] a co-decision of Commission, European Parliament and Council.”
Carsten outlined the “active safety” technologies already in place and, “based on the evidence reviewed, the following measures were considered to be likely to be cost-beneficial and could on that basis be taken into consideration: enhanced AEB with collision mitigation; intelligent speed adaptation; lane keep assist; reversing detection and reversing camera systems; and emergency brake light displays.”
Things are happening fast. “New vehicles sold from 2022 onwards are likely to have Intelligent Speed Assistance as a required fitment,” Carsten told TISPOL. However, “it remains to be seen exactly what type of system the Commission will propose. Hopefully there will be a dramatic increase in speed compliance and less work for the police”.
He remains hopeful that autonomous vehicles could reduce the number of accidents caused by people driving under the influence of alcohol or drugs. “The European Commission estimates that across the EU around 25% of all road deaths are alcohol-related,” he said.
“There were an estimated 240 fatalities in Great Britain as a result of drink-drive crashes in 2013, which equals about 14% of all fatalities [and] in Great Britain during 2014, 0.9% of drivers admitted to driving under the influence of illegal drugs at least once in the previous 12 months.”
Of course, he added, “automated systems do not fall asleep” and “25% of all fatal crashes on motorways and major trunk roads in the UK are sleep-related. In Victoria [Australia], around 20% of all fatal road crashes involve driver fatigue. For Western Australia, it is 30%.”
However, Carsten warned that people “will tend to be more sleepy in highly automated vehicles”. He cannot see autonomous vehicles having any beneficial effects if the systems are based on rapid human intervention in a crisis.
So, for Carsten, one of the biggest questions that the industry needs to answer is “how to manage the handover from automated back to human driving. It is one of the major design challenges,” he said. For instance, “how will the vehicle know that the human is ‘available’ to drive?”
The occupants might not be asleep, but they also might not be paying attention. They could be taking part in “a non-driving task such as video or email. The typical answer is to monitor the human but eye movement cameras are not 100% reliable and don’t work at all for some drivers.”
Carsten urged for a greater understanding of how the autonomous vehicle and its occupants collaborate. “They will need to understand each other’s authority, capabilities and intentions. The human machine interface (HMI) has a crucial role in communicating between the two,” but the perfect system does not yet exist.
Carsten cited a recent fatal crash in the US where the driver of a
“The Tesla driver was on autopilot and reportedly driving well over the speed limit. The truck driver did not detect the Tesla or thought he had sufficient time to manoeuvre. The Tesla driver was reportedly watching a movie. The forward vision sensor system on the Tesla reportedly did not detect the truck trailer, which was white against a bright background. The sensor system saw the space under the trailer as empty space (and carried on as if it was driving under a bridge or a gantry).”
As a result the Tesla collided with the side of the trailer at high speed, slicing off the top of the car.
So where might the blame lie in situations such as the Tesla accident? Could it be a sensor failure? Might there be too much vehicle speed? Would the road layout be to blame? Carsten said that the safety lobby is going to need to work on things like: “the speed limit for automated driving; better sensors; requiring side under-ride guards on truck trailers; geofencing … which means preventing the enabling of automated driving on lower-quality roads through use of a digital road map; previewing intersections via a digital map and disabling of automated driving on approach to an intersection; the control of distracting activities where attention is required; and vehicle-to-vehicle communication.”
Carsten said he could also envisage a requirement for an event data recorder, commonly called a black box and similar to ones in airplanes that track the performance of the aircraft.
In the long run, though, autonomous vehicles could be good news. “Driver assistance systems are likely to increase driver rule compliance and automated vehicles will be generally safer,” Carsten concluded “However, there are some real challenges in ensuring that safety is maximised and analysis of electronic data will play a major role in crash investigation.”
Distraction daze
The modern driver is facing an “epidemic of distraction” thanks to mobile phones, tablets, music systems and GPS screens. It is harder for a driver to focus on the road ahead, which is why sergeant Neil Dewson-Smyth of UK’s Cheshire county police launched the ‘don’t stream and drive’ initiative. He is using social medial to encourage drivers to keep their eyes on the road and their hands on the wheel.
“A well-trained driver and a vigilant car is a win-win scenario for the future,” said Neil Greig, director of policy and research at the UK’s Institute of Motorists. But he too warned that a plethora of devices hitting the market is creating a “Wild West in mobility” for users. The institute is also worried about driver awareness versus the car’s level of automation. Who has final responsibility for what? What activities unrelated to driving should be allowed? Where will lie liability and burden of proof in conditional and high automation?
Greig suggested that the car of the future should be benchmarked against the best human drivers. “This is a once-in-a-generation opportunity to ensure that technology replicates the best that humans can be. Let’s aim for 5-star cars driving on 5-star roads controlled by 5-star drivers.”
The more likely outcome, he warned, will be autonomous driving system manufacturers rushing to market with a mish-mash of unproven products. The market could be headed towards a battle similar to that of the VHS versus Betamax struggle for video recording supremacy, he warned. “Consumers will be left in the middle, not knowing who to trust, what to buy and which platform will become the industry standard.”
