Crash-proof cars!

I'm fortunate enough to be involved with various groups through my work with the Ergonomics Society, one of which is the Parliamentary Advisory Council for Transport Safety (PACTS), as I sit on their Vehicle Design Working Party. At our meeting this week, which was kindly hosted by Thatcham, we got to play with a bunch of new cars fitted with various advanced driver assistance systems.

I'm not going to risk my neck by naming manufacturers, but in short we were able to try a variety of collision mitigation and avoidance systems, as well as different variations on lane departure warning and lane-keeping assist systems. These particularly appealed to me as I'm interested in the effects of such automation on the driver and how the technologies interface with the human.

I've written a paper on how automotive automation can be divided into 'vehicle automation' and 'driving automation' - with the former being all those 'below the line' aspects, which don't impinge on the driver's conscious awareness (such as ABS or ESC), while the latter affect the more 'in your face' aspects of driving (such as ACC or lane-keeping systems). Collision mitigation is an awkward one as it can straddle the line - but in the cars we tried this week, the system only kicks in as a very last-minute attempt to save the day, way after the driver would respond, so in my opinion this falls just on the side of vehicle automation. So it was great to get the chance to try both sorts - and I'd like to share my own personal views on these.

On the whole, the collision mitigation systems were impressive, with successive generations improving on the last (and even a new prototype system that isn't on the market yet, which was widely viewed to be the best so far). Some of them had early warnings, like an icon on the dashboard or an auditory warning - which were really quite ineffectual in such emergency situations. The latest version improves somewhat on these by flashing a massive red light (pretty much a brake light) on your windscreen slightly in advance of intervening itself - giving you at least half a chance to respond.

There was much more variation in application - and opinion - for the lane departure systems though. Different manufacturers opted for different feedback options, whether that's a seat rumble, a steering wheel vibration, or an auditory warning; one manufacturer goes so far as to 'help' the driver by providing gentle resistance to get you back in lane. Personally I felt the steering wheel vibration was most ergonomic, as it's almost exactly what you'd get on a white line rumble strip - so very naturalistic and in line with drivers' expectations. The seat vibration came second for me, with its advantage of being directional (left buttock for left lane excursion and vice versa). I liked the lane-keeping assist, but I thought it was rather subtle and could easily be ignored by most drivers (it feels a bit like a road camber). The auditory warning again was least effective, giving no naturalistic relation to the task nor direction information.

Interface issues aside, the really interesting thing I noticed was the level of debate around the lane-keeping assist/departure warning systems as compared to the collision mitigation systems, with the latter meeting with universal approval from the group. Possibly this might have something to do with the more consistent interfaces, but I think there's also something in the vehicle/driving automation distinction, being as the collision mitigation systems were essentially below the line - like I said, way beyond the point of recovery for human reactions. It really was a safety net. Lane-keeping, on the other hand, was seen as still the driver's domain, and a lot of people were frustrated at the tactile feedback systems - despite these ostensibly being the most 'ergonomic'.

Ultimately, if we are to go down the route of crash-proofing cars through technology (which seems to be the trend - EuroNCAP now assess cars according to their crash avoidance technologies such as ESC), I think we do have to pay attention to the driver/vehicle distinction with automation, and be a lot more careful about how we implement driver automation.