BMF Response to Department for Transport Consultation Document on the review of Legislation concerning driverless cars
The important thing to acknowledge is that the right balance must be struck. Whilst the BMF does not want to see draconian legislation or regulations applied to such a degree that vehicle development is inhibited, it is important that sufficient legislative control is exercised to protect the safety of other road users whilst this technology is being developed and undergoing trials.
The BMF strongly believes that lessons from the past should be called upon to address this subject, and also that appropriate existing legislation should be called upon to provide safeguards.
It is also vital that the net is cast as wide a possible to engage industry and type approval experts in framing the regulations under which trials of these vehicles take place.
This should include, but not be restricted to, Type Approval Authorities, Technical Services, Vehicle Manufacturers.
Whilst the BMF accepts that the purpose of this review is not to develop detailed Legislation covering these vehicles, it is vitally important that vehicles under trial are operated in as safe and controlled environment as possible, to ensure the safety of other road users, and to maintain confidence in the trial.
Looking forward the BMF considers it essential that discussion begins very soon to consider the issues surrounding the introduction of highly automated vehicles to the general traffic mix. Much of the publicity touts high automation as a safety feature but there is too little information so far on how the systems are built and tested and how they ‘fail safe’ to inspire real confidence in vulnerable road user groups. We have seen no proof that any autonomous vehicle can correctly identify a motorcycle and its rider, and have considerable doubts as to whether they have actually been tested for this.
There are several areas of vehicle legislation currently in force which should be applied to vehicles undergoing these trials.
The main area which has to be addressed is the issue of Electro Magnetic comparability, commonly referred to as EMC.
As Motor vehicles have become more complex and vehicle control has moved from mechanical to electronic means, the issue of EMC and the interaction on electronic systems has become much more important.
It also has to be accepted that knowledge about these systems, their interaction, and EMC, are restricted to a limited number of experts.
Therefore for these reasons it is vital that robust risk assessments and FMEA's are carried out, by engineers who completely understand the EMC issues surrounding the systems used by these vehicles, and complete disclosure takes place.
Lessons learned from the past:
Any lessons from past vehicle issues should be drawn upon, including those related to Electric Regenerative braking systems on hybrid vehicles and the interaction between such systems and foundation hydraulic braking systems, and ABS modulation.
Finally The BMF insists that before prototype cars with high automation are used on the Public highway they are fully tested for compliance against the legislation highlighted in our answers and Test Reports are issued by a recognised Technical Service.
This is vital if confidence in the project is to be maintained.
General comment on the discussion paper:
The BMF notes that the title of the paper uses the term ‘driverless cars’ but this term is apparently being to used to cover both vehicles with full automation (truly driverless) and vehicles with high automation but ultimately under human control. We consider that these need to be clearly differentiated as the liability for each seems to us to be different.
The questions appear to be about both the testing of the vehicles and the longer-term application; we have tried to answer appropriately but in the absence of an overall strategy with respect to highly automated vehicles there is too little information on what is to be tested.
Answers to specific Questions:
Q1.Should any special training/testing or a minimum number of years of driving experience be specified for drivers involved in testing driverless cars with high automation?
For testing purposes it would seem sensible to utilize drivers with a broad spectrum of age and driving experience. Initially very experienced drivers with training and experience above and beyond the DSA standard. As the learning takes place this would aid these lead drivers to subsequently act as observers for drivers of less experience, perhaps with limited dual-controls.
Q2. Should a second person be required to be present, as an observer?
Yes. This will assist the driver as well as serving as support for additional drivers. In addition to this it would seem sensible to use an observer to capture the experiences and outcomes of each drive.
Q3. Do you believe that the normal set of requirements for driver behaviour should still apply or are any exemptions from these required, if so please specify?
In the text there is a suggestion that the decision to pursue a charge of Careless driving may arise if the behaviour of the vehicle (“…….vehicle is wandering across the road……”). This suggests that this can happen. If this can possibly occur, then the driver must be attentive and thus ultimately deemed to be in control of the vehicle. Therefore having hands at the controls must remain a requirement. A wandering vehicle could equally equate to a vehicle failing to avoid another or to stop.
In a car with full automation the situation is different, as the driver does not exist; no-one can intervene and thus all in the vehicle must be deemed to be passengers. The vehicle manufacturer must hold responsibility and insure accordingly.
Exemptions must not be made in the earlier phases of testing, the normal set of requirements for driver behaviour should still apply.
The BMF is concerned that in the longer term high automation leads to dumbing down the skills of car drivers. In the case of autonomous vehicles it could be that no skill at all is required. The sooner the implications of this are addressed the better.
Q4. Are any new requirements or constraints necessary?
With respect to drivers, no. In the long term, if truly driverless cars do become part of the traffic mix it would be necessary to drop the driver behaviour requirements. But this cannot be allowed to happen unless full liability is accepted by vehicle manufactures.
Q5. Do you have any suggestions for an indication to other road users that the vehicle is operating autonomously, or capable of autonomous operation?
On one hand there is no point in indicating to other road users that the car is being driven by a machine with a person in position to override its decisions because no one will understand what to expect or what is expected of them. On the other hand the other road users must be made aware of an experimental vehicle on the road. There are too many vehicles on the roads equipped with flashing beacons and other road users seldom understand the reasons why any one of them is so equipped. The BMF suggests that the vehicles on test be painted in a highly conspicuous colour and marked on the bonnet, sides and rear as “Experimental vehicle”. For night time operation this could be augmented with low powered steady lighting strips.
Q6. Should educational materials be developed to advise other road users about the testing of highly autonomous cars?
Yes. Nationally and to a very high level in the cities where these trials are to take place via a media campaign.
Q7. Do you have any observations on the possible reactions of other road users, or the risks of interaction with driverless cars and possible mitigation measures?
Uncertain, anxious, worried and potentially frightened. The three vulnerable road user groups will be most concerned and the mobility restricted, elderly and parents of young children are likely to be particularly concerned. The BMF represents one such group and has profound doubts.
We also have concerns relating to the interaction of highly automated cars with the all vehicles/road users not so equipped; for example, it may be possible to implement a fully automated braking system which can outperform any human driver. In some circumstances this might actually cause an accident rather than preventing it, if the following car cannot match the performance.
Q8. Do you see any difficulties with the existing product liability regime, when operating driverless cars with high automation?
Yes. Like a number of the other questions this is ambiguous as to whether it relates to High Automation or Full Automation. In the case of High Automation where driver intervention is possible the BMF sees potential for the development of a whole new legal minefield as the insurers of the drivers and vehicle manufacturers dispute liability in an ocean of expert witness based claims and defences before bewildered Judges. This needs to be pre-empted by discussions starting very soon and including all road user groups in the interested parties.
Q9. Do you have any suggestions for standards to regulate the testing of prototype cars with high automation?
Clearly these vehicles will involve a high level of electronic control and the issue of Electro magnetic compatibility is key (EMC).
Therefore all vehicles undergoing trial must comply with UNECE Regulation 10 in its latest form and full disclosure of the systems operation should take place prior to a complete risk assessment of the systems used. This must include a full FMEA.
Vehicles undergoing trials must be tested for compliance with UNECE Regulation 13H including all aspects of the sections dealing with ” Special requirements to be applied to the safety aspects of complex electronic vehicle control systems”. See Annex 8.
Vehicles undergoing these trials must comply with UNECE Regulation 79 in it's latest form, including Annex 6 “SPECIAL REQUIREMENTS TO BE APPLIED TO THE SAFETY ASPECTS OF COMPLEX ELECTRONIC VEHICLE CONTROL SYSTEMS”
Vehicles undergoing test MUST comply with these requirements, and this should include test reports from recognised Technical Services confirming compliance, prior to any vehicle being used on the public highway.
Q10. Are there current type approval or construction rules that prototype cars with high automation might not comply with?
This is a very difficult question for anyone to answer without access to said vehicle, clearly vehicles must comply with the legislation outlined above. It is hard to envisage trials being allowed whereby vehicles do not meet minimum requirements for steering effort, turning circle, braking performance and EMC.
Q11. Are you able to suggest any specific areas (e.g. braking, steering) or any specific systems/technologies (e.g. ABS, ESC) where regulation needs to be amended or developed, as a priority?
Again this is a difficult question to answer without access to said vehicles, however the answers supplied in questions 9 and 10, should act as a starting point.
The BMF, as a group representing vulnerable road users, requests that they are involved in the development of such legislation, particularly when it is remembered that modern motorcycles incorporate many electronic systems such as engine management ECUs, “fly by wire” throttle control, and ABS.
Any uncontrolled interaction between these systems and those deployed on prototype cars with high automation, must be considered and addressed prior to any trial taking place.
Q12. Are any changes to the current roadworthiness regime required to permit the testing of driverless cars or ensure their safety?
For testing purposes the experimental vehicles should initially be subject to a roadworthiness test at least equivalent to the MOT prior to every test drive. With respect to the automated systems it should be a given that they self-test on start-up. As we have no information on what is actually being tested we cannot comment further.
Q13. Have you any initial thoughts about any longer term risks and issues as driverless cars age, and possible requirements to address this?
The BMF is glad to see that in this section someone in the Department has had some thoughts about the longer term implications and will be happy to participate in future discussions on vehicle longevity. As we currently have no knowledge of the durability and fail-safe aspects of many of these systems we cannot usefully comment at this stage. We do have concerns that for the manufacturers of cars and the carriers of the products insurance/liability there will be a commercial incentive to withdraw support for models at a certain age thus introducing the prospect of absolute designed obsolescence on the manufacturer’s timescale.
Q14. Re vehicle tax, registering with DVLA, comments.
In the longer term it is probably desirable to have the level of automation on the car made part of the vehicle details, just as e.g. engine capacity is now.
If the use of CO2 emissions is deemed an appropriate means of assigning VED currently then it should be just the same for these vehicles in any case.
Q15. Do you anticipate a need for special infrastructure to permit the testing of cars with high automation?
No. Certainly not initially as such vehicles must be capable of using the existing infrastructure and, most importantly, capable of interacting in a safe manner with all other road user groups. The use of highly automated vehicles must not preclude the use of vehicles not so equipped on the same roads.
If substantial or even relatively minor amendments of the road networks are to be a requirement of these vehicles being used on the roads then we would suggest that the manufacturers of the machines be asked to review their designs. The UK network is vast and any alterations will place a substantial burden on the taxpayer. In addition they will most likely deflect funds for general capital works as well as introducing a substantial ongoing burden on revenue budgets for the maintenance of these modifications. It is these very budgets which have been most severely constrained during government reductions in spending. Historically it is this area which always suffers. The costs of the projects should be borne initially by the motor manufacturers and passed on to the product purchasers, the beneficiaries.
We have no knowledge as to whether any of the machines to be tested imminently require this but are aware that there are proposals for V2I systems. Vehicles so equipped must also be capable of operation without the specific infrastructure.
Q16. What issues would need to be addressed to enable insurers to offer suitable insurance products?
Is this relevant to the initial testing?
Q17. Are there other insurance related issues which may affect the introduction and testing of driverless cars?
The BMF considers that discussions on issues of liability should start as soon as possible.
In the case of fully automated vehicles we consider it appropriate that that full liability rests with the vehicle manufacturer. In the case of highly automated vehicles, where a human driver can intervene, there is the possibility of a whole new legal minefield where the insurers of the drivers and vehicle manufacturers dispute liability in an ocean of expert witness based claims and defences before bewildered Judges. It may be possible to pre-empt some issues with additional requirements for in-vehicle systems, e.g. logging human driver input.
Q18. Do you have any suggestions or concerns over data collection and privacy when considering the testing of cars with high automation?
The BMF has no concerns about data collection and privacy during testing; we consider that the data should be public as possible. If/when highly/fully automated cars come into general use we would suggest that e.g. location data were automatically deleted after say 30 days except in the event of a traffic collision.
Q19. Do you (a) support amending diverse current regulations to cater for driverless cars alongside conventional ones, or (b) support creating a special regime via specific regulations to permit the testing of driverless cars under certain circumstances or constraints?
Another question which is ambiguous as to whether it relates to High Automation or Full Automation.
In the case of High Automation, if the driver can intervene there seems to be little reason to alter present driver requirements or regulations. In the case of Full Automation almost everything will have to be reviewed; the BMF is concerned that the driver of the conventional car or the member of one of the three vulnerable road user groups may be always deemed to have been at fault because Judges are unlikely to be able to resist the expert witness pressures applied by manufacturers.
Q20. Do have any other comments on the need for a special regime to cover the testing of driverless cars with high automation? Do you consider any other regulations or aspects of driving practice would pose a barrier or do you consider that extra conditions would need to be imposed?
This is a new area we are entering and the linking of existing electronic vehicle control systems was in some respects inevitable, and the days of complete mechanical control of road vehicles are behind us.
However the BMF strongly urges caution and urges all involved to accept that complete knowledge of the uncontrolled interaction between electronic vehicle systems is restricted to a few experts in the field.
Unexpected vehicle events have occurred in the past due to EMC issues and everything possible must be done to, draw on the lessons learnt from these experiences to ensure safety of all road users, whilst these trials take place.
There are a number of issues which will have to be dealt with beyond this initial testing phase, interaction with all other road users, liability and insurance issues, vehicle longevity, changes to driver testing, system security to name just a few, in addition to the EMC issue above. We look forward to participating in such discussion with a view to making this work for all road users, regardless of their chosen mode or their age or the age of the vehicle.