Uber gyrometer monitoring is the groundwork for the future of car insurance
Tue 26 Jan 2016
Ride-sharing giant Uber is running trials to develop procedures that allow the company to verify negative customer feedback by measuring data from gyrometers in drivers’ smartphones – an initiative which seems certain to be closely followed by the motor insurance industry, which is increasingly moving towards intensive monitoring of driving behaviour as a factor in costing insurance premiums.
The announcement was made today by Uber’s chief security officer Joe Sullivan, who outlines the pilot scheme as an investigatory measure rather than a real-time monitoring scenario. At the end of an Uber ride, both the driver and the passenger/s are obliged to rate each other using the Uber app. If an unusually low rating is given by the customer under the trial scheme, any feedback that can be verified by examination of the journey data, such as speeding or otherwise erratic driving, will be investigated using data from the gyrometer in the driver’s smartphone, together with data from GPS and accelerometers. Sullivan writes:
“Gyrometers in phones can measure small movements, while GPS and accelerometers show how often a vehicle starts and stops, as well as its overall speed. If a rider complains that a driver accelerated too fast and broke too hard, we can review that trip using data. If the feedback is accurate, then we can get in touch with the driver. And if it’s not, we could use the information to make sure a driver’s rating isn’t affected.”
Gyrometer data can provide evidence of sudden jolts, tailgating stops or jarring transit for other reasons, whilst accelerometer data can be used not only to evaluate driver performance on a particular journey, according to Sullivan, but also to develop ‘mean averages’ for established routes, such as from a city to an airport. With benchmarks established, deviating metrics such as above-average speed for that route, can be utilised for quality control purposes.
The trial began for select Uber drivers in Houston in late 2015, and the company has also revealed that it is planning further unspecified experiments to develop driver monitoring techniques in order to improve safety.
Sullivan notes in the announcement that speeding caused 1 in 3 fatal accidents in the United States in 2012, resulting in 10,000 fatalities that year. ‘And distracted driving,’ he writes. ‘is just as big a problem. In 2013, over 3,000 people were killed in car crashes that involved distracted drivers. Texting or talking on the phone triples your chances of getting into an accident.’
Though not explicitly mentioned in details of the pilot scheme, this mention of talking on the phone whilst driving seems to indicate that it is either already a factor in this or a similar unannounced Uber experiment, or that it is likely to become one soon.
In 2012 insurance company Aviva was among the first to develop smartphone-based metrics which did not require any company-provided telematics device, and offered 20% discounts to customers who participated in the scheme. The Aviva ‘Rate My Drive’ scheme developed driver profiles – and risk assessments – based on 200 miles of average driving.
Progressive Insurance, the biggest auto insurer in the U.S. that is actively pursuing driver monitoring, also offers 10-15% discounts for drivers who will submit to the same cocktail of monitoring which Uber drivers must agree to in the new gyrometer trials.
In the UK and Italy between 15-18% of car insurance policies are based on telematics, and consultancy service Ptolemus predicts [PDF] that up to 40% of UK policies will involve telematics by 2016.
It’s worth noting that the reductions in policy premiums to early smartphone-monitoring adopters seem to be dwindling with the years, and with the increasing acceptance that micro-monitoring of drivers is to become the norm, with many predicting that those wishing to ‘opt out’ of this level of surveillance will initially be punished with higher premiums or the need to resort to ‘specialty’ insurers – and eventually will be unable to obtain insurance at all.
Uber’s monitoring trial involves accessing data only when a customer has flagged up errant behaviour, and the reasons for this seem to be more infrastructural than expressive of any particular concern for privacy. The volume of data produced by anyone who drives for a living is considerable. Should it all be stored to the cloud automatically on a live basis, or stockpiled in relatively low-capacity local storage until a driver emerges from a connectivity blind spot? At what point do you ‘clear the buffers’, considering that police and other authorities are going to be very interested in the prospect of accessing a kind of automotive Akashic record of every move every driver makes, presumably with the purpose of crime investigation and evidence-gathering for other purposes?
Network limitations seem likely to throttle the IoT-style dream of full-time, real-time vehicle behaviour analysis, if that prospect wasn’t almost equally crippled by the demanding compute cycles and AI advances that would be necessary to provide a ‘live map’ with granular data of all ‘connected’ motor traffic over a large swathe of geography.
However these are temporary bottlenecks – the trials Uber are currently carrying out do seem to represent the future of motoring.