Insuring the heterogeneous vehicle population comprising automated and manually driven vehicles
04 Feb 2015
The global automotive industry aims to satisfy three primary goals – saving the environment, saving lives and saving effort in commuting. These three goals can be realised through intelligent mobility, the convergence of automated driving and cooperative driving – which will provide the basis for setting future insurance premiums.
Active safety features such as active lane-keeping, autonomous emergency braking, adaptive cruise control with stop-and-go etc. form the building blocks of automated driving. To enable these technologies to fully realise their safe driving potential, vehicle-to-vehicle (v2v) and vehicle-to-infrastructure (v2i) communication is required – and these are the building blocks of cooperative driving. Neither automated driving nor cooperative driving are independently and separately capable of enabling perfect unmanned driving manoeuvres. Cooperative driving is important, for instance, in understanding the intended driving trajectories of other vehicles so as to make safe lane-changes or cornering manoeuvres. Similarly, v2v and v2i communications are not capable of detecting animals or obstacles ahead without using some sort of Advanced Driver Assistance Systems (ADAS). The convergence of automated and cooperative technologies compensate for each other’s flaws resulting in a near-perfect driving. Smarter driving environments, such as smart traffic signals that communicate with the vehicles, smart road lanes that provide the right turn-radius, smart buildings that enable unmanned parking etc. will allow cooperative driving and automated driving can be done in a much more efficient manner, further enabling intelligent mobility.
Intelligent mobility is an enabler for electric mobility as well. Human drivers may forget to charge their electric cars often. However, intelligent vehicles will learn to charge themselves. Intelligent vehicles will learn to drive closer to each other, compared to the inter-vehicular gap maintained by human drivers as these vehicles will know the precise braking effort needed to ensure a safe-stop. This technology should go some way to alleviating congestion.
As these vehicles lack the human aspects of error such as drowsiness, recklessness, distraction, urgency and frustration, the intelligent vehicles are unlikely to cause crashes. This has major implications for several market participants such as body-material manufacturers, glass-manufacturers and various other traditional suppliers. One of the biggest impacts is for the automotive insurance industry. If vehicles are capable of driving such that 90% of accidents, which could be attributed to human factors, are averted, the frequency of claims for collision-related cases will plummet. This would result in the vehicles warranting lesser insurance as the risk is minimised. The reduction in the insurance premiums would be so severe that the fragmentation of the automotive insurer-base would no longer make sense. Some auto insurers may go out of business, some may acquire others and some others may choose to partner with other ecosystem participants such as warranty & after-sales service providers to create a bundle of services. Simply put, auto insurance, as we know it, will undergo dramatic change once the entire traffic is comprised of automated driving vehicles.
The first fully-automated vehicles are expected by around 2020. It would be a good 20 years before these vehicles form a sizeable portion of the market. Consider electronic stability control (ESC), a life-saving technology that minimizes wheel “skid” while turning. This technology suffered poor take-up rates in Europe and in North America until legislation was introduced requiring all new vehicles to be fitted with ESC. After almost 20 years since market launch, there were still more than 25% of new vehicles sold in 2012 that did not fit this technology. It is not until 2020 that we expect Europe to have all new vehicles to be fitted with ESC, despite legislative push. It will be therefore almost 25 to 30 years since introduction and a strong legislative push that can drive a 100% new vehicles fitment scenario. Even then, existing vehicles need not be retrofitted. Hence, for all the vehicles in operation to be fitted with ESC, it would take another 15 years, given the average replacement rate of vehicles. This suggests that almost 40 years might be required from the introduction of a technology to making sure it is universally present across the fleet.
A chain is only as strong as its weakest link. The weakest link in the intelligent mobility environment is an unintelligent vehicle that is manually driven. One manually driven vehicle that collides with one or more smart vehicles would result in a domino effect in traffic, causing at least traffic congestion in the best-case scenario and loss of lives in the worst-case scenario.
There are several ways of determining risks of a heterogeneous traffic that comprises both manually driven and automated vehicles. From an approach similar to marginal costing, we could estimate the impact of one lesser manually-driven vehicle or the impact of replacing one risky vehicle with one minimal-risk vehicle, using highly sophisticated computer models. Yet, this approach would not fully-well take into account the fact that some “automated” vehicles come with manual driving modes as well. Therefore, it is not the traffic, but the trips of an individual vehicle that needs to be considered for insurance risk calculation.
Similar to the financial analysis industry, we could adopt a 'trailing twenty hours' approach to increase or decrease the insurance premium calculation, resulting in a by-the-last-hour up-to-date insurance risk applicable to the vehicle. Every manually driven mile needs to be billed at a different rate against every mile driven in the automated mode. However, not every manually driven mile is same across various drivers. Bring in the pay-how-you-drive aspect and innovative measurement modes employed by companies like The Floow, as adopted by the Direct Line Group, we can solve one main aspect – the insurance aspect of the manually driven miles. We are still left the automated driving paradox.
The software and the hard-wear
Not all automated driving vehicles are created equally. Car manufacturers will want to differentiate themselves. Some cars will want to remain 'sporty', suggesting a less risk-adverse autonomous driving mechanism than more staid and steady models. Car owners may not be content with accepting autonomous technology as a given – they may try to tweak and customise their vehicles, even to the extent of influencing underlying algorithms. This clearly puts those cars in a separate risk category; potentially much riskier, depending on the technical expertise of the tweaker.
Assuming no human interference, different algorithms remain difficult to assess. Insurers and underwriters need to be able to test the vehicle’s driving algorithm under different driving scenarios, across various driving conditions. The impact of wear-and-tear on the reliability of the driving algorithm is still questionable. A showroom-condition vehicle may produce certain results in test. The same vehicle with worn-out, under-inflated tyres and worn condition brake-pads will not perform with anything like the same level of efficiency. To ensure the algorithm performs adequately, the physical hardware of the car need to be kept in something close to its original condition. That suggests a regular servicing requirement, perhaps as often as every 100 miles. Cars that can perform with minimal wear and tear will be at an advantage. Preventing wear and tear is as much a product of driving style as it is of the hardware itself; suggesting another advantage to the steady over the sporty models.
Hardware and software issues aside, there are a number of questions to the driving of a vehicle which insurers will have to understand. These include: how many modes of automation does a vehicle possess across semi-, highly- and fully-automated modes? What is the time given to the driver to take-over the driving task? How many driver-distraction inducing features are fitted in the vehicle? What is the safe-stop manoeuvre or minimal-risk manoeuvre that a vehicle must adopt in case the driver fails to take over? What is the guarantee that the vehicle will complete the safe-stop or minimal-risk manoeuvres without resulting in any damage to itself or surroundings? Each of these aspects can be similarly analysed and quantified to come up with a range of minimum and maximum values of insurance premium – which are expected to be less than that of the insurance premium for completely manually driven unassisted vehicles, as the impact of human error is minimised.
Feasible approaches for calculation of insurance risk will be required for vehicles that can be driven both manually and automated. However, depending on an underwriter’s level of belief in the driving algorithm, the value of the premium to be paid by the automaker would vary. In all probability, no automaker will empower their top-of-the-line large-sedan with the same driving algorithm as installed on their entry-level small-car. Bigger cars should be capable of more advanced manoeuvres is how the industry views the vehicle segmentation. Therefore, there could be a differential or preferential range of insurance premiums applicable to different vehicles of different makes, across different segments.
In essence, it is imperative to understand and quantify the number of manually driven and automated driving vehicles that surround a vehicle in order to be able to calculate the right insurance premium in a heterogeneous traffic mix.
Download full article (PDF 149 KB)