29th May 2019
Ross Clarke, a chartered engineer and road traffic collision investigator from Hawkins’ London Office, recently gave a webinar to the Forum of Insurance Lawyers (FOIL) on one of the most thought-provoking emerging technologies in the automotive field: autonomous driving.
Ross notes that it is important to distinguish that automated driving and autonomous driving are very different. Currently there is no autonomous vehicle readily available to the general market, and development of this technology is still ongoing. An autonomous vehicle, by definition, would not require any level of human interaction, while a partially, or conditionally, automated vehicle would still require some human input in order to operate.
The scale of automation based on a driver’s control of a vehicle. (Source: The United States Department of Transportation’s National Highway Traffic Safety Administration)
Though there is currently no defined universal procedure for how forensic investigators will extract data from autonomous vehicles, Ross speculates that the use of current technology like Event Data Recorders (EDRs) could continue. EDRs on some vehicles already log information around the time of an incident, such as the vehicle’s speed, the application of controls, and the vehicle’s occupancy. Most modern vehicles in Europe include some form of EDR, although an EDR’s main purpose is to supply the vehicle’s manufacturer with critical information relating to the car’s safety systems. Access to EDR data is usually restricted, and is currently only provided to forensic investigators voluntarily, and only by a limited number of manufacturers in the UK. Ross believes that for autonomous vehicles, EDRs will probably perform the same functions, but he expects that the amount of information stored in relation to each crash event will greatly increase.
Benefits of autonomous driving would be safety, productivity, mobility, cost, and reductions in both traffic and emissions. The ability to remove a human driver would not only afford increased independence for those who are unable to drive, but will also benefit other drivers on the road by removing human error associated with them. Ross notes that in 2016, 94% of all serious collisions in the United States were linked to human choices.
Risks involved in autonomous driving, however, would centre primarily around privacy and cyber security, due to the vehicle’s controls and the stored personal information. For example, autonomous vehicles need to map their surroundings in order to understand their environment, but mapping of private property could be considered an intrusion. There have, unfortunately, already been incidents involving vehicles using some form of autonomous driving feature, which can often happen during the testing and introductory phases of new technology. One such incident occurred when a vehicle collided with a pedestrian walking her bicycle across the road; the vehicle was unable to recognise the pedestrian from its database of distinguished objects, and therefore did not determine the need to brake. It is clear to developers of this technology that simply identifying the location of objects is not enough: the vehicle must also distinguish what each object is, as well as its likely trajectory.
Current autonomous vehicles are nearest to Level 3: Conditional Automation. This is a huge step for vehicle technology, as the driver would essentially be a backup rather than the primary control system in this scenario. A number of pieces of emerging technology and software would now come between the driver and the vehicle, which will lead to discussion surrounding who, or what, is potentially liable in the event of a collision.