Ljubo Vlacic1 Jul 2016
Please note that this submission is my personal view and cannot be interpreted as the view of Griffith University.
I am of the view that SAE International Standard J3016 regarding levels of driving automation should not be used as part of the legislation, not for the time being at least. This is because:
a) The technology development is progressing at its own pace and much faster than the pace of change of the legislation needed to reflect advancement in the technology deployment.
b) At this transition stage the automated vehicle developers have full freedom in decision making about the level of automation they wish to assign to their vehicles and, moreover, freedom in making decisions about way of deploying automated driving functions and, thus, interactions between the vehicle and a human being inside the vehicle.
c) The Discussion Paper (May 2016), being designed around the SAE Standard and in attempt to describe a driving situation for each automation level, appeared slightly complicated and, to some extent, inconsistent with respect to automated driving functions the document is making a reference to.
d) Without having an event recording device installed in the car it would be very hard to check, after the accident, whether a human driver timely received “hands-on” signal from the vehicle; moreover, it would be even more difficult to check whether the driver correctly and timely responded on to such a warning signal received from the vehicle. Perhaps, we may be able to consider regulating driver’s intention and variety of interfaces and interactions between a human driver and automated vehicles at some later stage of the technology development.
e) The current formulations, offered by the SAE Standard, are not precisely correct, strictly technically speaking, and thus need to be further refined.
The most that should be done is to include the SAE Standard in the appendix of the document which will be submitted to the Transport and Infrastructure Council for the purpose of assisting the Council in understanding automated vehicle functions and offering an example on how they can be performed.
My view is that the question on whether the vehicle is equipped with a steering wheel, or not, should be used as the only point of distinction in answering the question on whether the vehicle was operated by a human being or by the vehicle in its own right. Therefore, as long as the vehicle is equipped with a steering wheel, the vehicle is assumed to be operated (or overseen) by a human being regardless of the level of automation made available to the human driver to use. This would mean that only two cases should be recognised and addressed by the legislation at this stage, namely:
(i) a vehicle without steering wheel – i.e. fully automated vehicle; and
(ii) a vehicle with steering wheel – i.e. partially automated vehicles which can be equipped with a varying number of vehicle monitoring and automated driving functions.
Speaking in terms of the legislation, the human driver is responsible for an accident made by a partially automated vehicle while the car designer & manufacturer should be responsible for an accident made by the fully automated vehicle.
In this context, for example, the definition of the Google Vehicle, page 33 of the current document, cannot be defined as a fully automated vehicle as it still has to be driven by a human being if the Google vehicle is on roads that have not been mapped. Therefore, the driver is still responsible for any accident caused by the Google vehicle from the viewpoint of the steering wheel-based vehicle classification regardless whether the accident occurred on the road that was mapped or not mapped. This in particular as a GPS signal can be lost even if the road was mapped. So, again, a black box (an event recording device) would need to be installed to enable “after the accident” investigation.
If this suggestion is accepted then the current document can be simplified significantly.
a) Thus, there will be no need to include in legislation fine details of automated driving functions (Chapter 3) apart from splitting them into two main categories, namely:
a. fully automated; and
b. partially automated driving functions.
Thus, the majority of the current content of Chapter 3 can be moved to the appendix of the document.
b) The same should relate to behavioural competencies and human drivers should be asked to demonstrate that they are fit to decide about safely deploying variety of automated vehicle functions should they wish to drive partially automated vehicles. This means that a new regulation would need to be introduced aimed at asking drivers of partially automated vehicles to undertake a training courses. It should be noted that, at this point in time, car manufacturers are only allowing their well - trained drivers to oversee on-road trials of their fully/partially automated vehicles.
c) My answers to the specific questions listed on page 28 would then be as follows:
a. Q1a: Yes, I do.
b. Q1b: All trials should meet the current road safety regulations. No exemptions should be in place. The road safety should not be compromised.
c. Q2a: It is now easy to answer the question as to who is in control of the vehicle, namely: (i) human driver is in control of a partially automated vehicle, and (ii) human being is not in control of a fully automated vehicle as the fully automated vehicle does not have a human driver
d. Q2b: There is no a need to define a new term “proper control”
e. Q3a: Government should only oversee the safe operation of fully automated vehicles. With respect to safe operation of partially automated vehicles, Government should consider introducing a provision for human drivers to keep demonstrating their fitness in deploying varying number and levels of monitoring & automated driving functions while driving partially automated vehicles. Manufacturers should be allowed to introduce any advanced function of their choice but human drivers should be responsible for making decision to use and deploy them.
f. Q3b: It should all be about safe driving
g. Q4a: A human being is a driver of partially automated vehicle while no human driver is in existence in a fully automated vehicle
h. Q4b: Human driver is responsible for driving partially automated vehicles regardless of the level of automation that was in place at the time of the accident; the designer and/or manufacturer of the fully automated vehicle is 100% responsible for an accident caused by the fully automated vehicle.
i. Q4c: I would leave this up to the lawyers to elaborate; the current chapter 7 can be significantly simplified
j. Q5: A human driver should be deemed responsible for the actions of the partially automated vehicle. The registered owner (or the vehicle designer & manufacturer) should be deemed responsible for the actions of the fully automated vehicles. The current chapter 8 can be significantly simplified.
k. Q6: No vehicle standard exemptions should be in place as, otherwise, road safety would be compromised. However, the national consistency of vehicle standards should only be required for fully automated vehicles. To this end it should be important to address the operational reliability of fully automated vehicles which, strictly technically speaking, relates to the issue of operational redundancy. Regarding partially automated vehicles, it would be up to the market (i.e.human drivers) to decide whether to buy them. The point is that, strictly technically speaking, partially automated vehicles can be used in various jurisdictions providing that vehicle-to-vehicle and vehicle-to-infrastructure protocols are consistent throughout the country. However, should Australia wish to ensure that only the top of the driving technology is imported by Australia then the national consistency of vehicle standards should be in place. Both, the fully automated vehicle and partially automated vehicle must have installed an event recording device (black box).
l. Q7: Issues related to vehicle modification and in-service compliance during the vehicle warranty period should be addressed by industry-led standards. Otherwise, any period beyond the warranty term should be addressed by the local legislation.
m. Q8: The liability acceptance becomes significantly simplified if my suggestion about the steering wheel-based vehicle classification is accepted.
n. Q9: I think, the data (information) ownership matter should be addressed first. If so, then the privacy information issue will become much easier to resolve. Namely, my view is that data should be owned by the entity that generates it. Thus, (i) data generated by a road side unit should be owned by the road transport operator; (ii) data generated by partially automated vehicle should be owned by its owner (which typically is not the car manufacturer); (iii) data generated by fully automated vehicle is owned by the vehicle designer/manufacturer; (iv) data generated by a transport service provider should be owned by the transport service provider; and (v) data generated by a human driver while driving partially automated vehicle should be owned by the human driver. Therefore, data generated by a human driver should continue to be regulated by privacy principles.
o. Q10: As per the discussion presented above, the only vehicle classification should be based on the ground whether or not the vehicle is equipped with a steering wheel assuming that a steering wheel is there for the purpose of enabling a human driver to take over the control of the vehicle at the point in time when she/he decides that it must be done.
p. Q11: As per the discussion presented above, an attempt to legislate varying levels of automation (except the main two ones – as discussed above) does not seem to me to be necessary. Moreover, it is not feasible as no car manufacturer is installing an event recording device in their automated vehicles. In addition, as per my comments to Q6 above, the operational reliability (and, thus, the vehicle architectural redundancy) is the issue which must be addressed by the regulator for the sake of both the transport safety and transport efficiency.
q. Q12: Yes, the staged approach is the way to proceed with.
r. I already presented some aspects of this my view during the consultation meeting in Brisbane on 06 June.
If NTC decides not to proceed with my main concept (that of making the distinction between vehicles based on whether a steering wheel is installed in the car or not) then I wish to suggest to NTC to consider:
a) Modifying current definitions of levels of automations by way of replacing the term “automated driving system” with the term “on-board vehicle navigation & control system” as a new technology has been developed which enables remote navigation and control of vehicle operations. This should be done not only on page 23 and page 31 but throughout the whole document.
b) Replacing the term “operating system” (page 23, line 1) to read “the on-board vehicle navigation & control system”
c) Modifying the current definition of the “automated driving system entity” as it may not be necessarily correct. In fact, the current definition would be deemed as incorrect from the viewpoint of the vehicle classification I proposed in this submission.
d) Modifying the current definition of C-ITS, page 23, as it is incorrect. Namely, C-ITS is about co-operation among traffic participants while the data exchange is just one of the enabling mechanisms for achieving it.
Of course, I would be happy to further elaborate on this matter if necessary.