My piece on driverless cars a month ago drew some attention and some excellent debate, although there was also some confusion about what “adaptive cruise control” actually is, and why it is in fact a fairly crude piece of technology (or, at least, will soon become so, in all likelihood).
Firstly, cruise control itself is not particularly well understood, for the simple reason we do not really think about it. The system works generally by the driver pressing a button or flicking at a particular speed, at which time the vehicle takes over the operation of the accelerator. This means the vehicle will maintain speed for as long as there is no intervention from the driver, nor anything which would require braking. In other words, the driver still has to look out for obstructions (including slower moving traffic), and notably also has to brake or use gearing on downhill sections to avoid speeding up (as the vehicle only controls the accelerator, nothing else). Primarily for safety reasons, this system was never available at below 40km/h or 25mph.
Secondly, the addition of a radar at the front of the vehicle to pick up traffic in front resulted in “adaptive cruise control”, which has been available on family cars for over a decade. However, in its initial phase, this was of limited additional value as the vehicle still controlled only the accelerator. On motorways it meant the vehicle itself could ease off the accelerator in line with the broad traffic flow (slowing gradually from around 100km/h or 60mph to around 80km/h or 50mph, for example), but still could not brake. The system also continued not to operate at all at low speed, thus had no value in queuing traffic and did not become widespread.
Thirdly, the addition of control of the brake pedal to the adaptive system led to the more modern “adaptive cruise control” (which really should be called something else, for clarity – perhaps “full range adaptive”). This means the vehicle takes control of the brake pedal as well as the accelerator, enabling: a) quicker deceleration when a vehicle is detected in front; b) maintenance of speed even on downhill sections; and c) reduction in speed below 40km/h, indeed even to zero. The system can thus be applied either at above 40km/h or whenever a vehicle is detected in front (even when both vehicles are stopped). Also, unlike regular cruise control, to system generally disallows acceleration until the vehicle is straightened up, enabling speed to be maintained through bends in line with preceding traffic, even if the vehicle disappears from the radar view on the bend itself (provided a speed of above 40km/h is maintained).
“Detection of a vehicle” is the key phrase, however. Modern adaptive cruise systems essentially identify objects in front as either “vehicles” or “obstacles”. Only objects moving forward in the same direction, or stopped directly in front when the operating vehicle is stopped, are deemed “vehicles”; anything else is deemed an “obstruction”. In the former case, the radar system will bring the speed into line (even down to or up from zero); however, in the latter case the driver is expected to take action – which is notable, because the latter case (the “obstructions”) even includes stopped vehicles when moving towards them (for example other vehicles stopped at lights).
The reason for this is that if these “obstructions” were defined any other way, you would find vehicles with adaptive cruise braking at in all kinds of odd places – for example when vehicles come the other way on curves or when road signs appear straight ahead as the road itself bends.
The challenge, therefore, is for technology to better define what is an “obstruction” and what is a “vehicle”. Attempts are being made at, for example, slowing the car for bends ahead (using satnav) or even scanning the scene in front to pick up pedestrians liable to walk out in front. However, although advances have been made, of course 100% assurance is needed before such systems can become widespread; and it should be noted that even modern adaptive cruise control causes some confusion with what it can and cannot do.
It is worth re-emphasising that I think we will see marked advances to the extent that even lower end cars a generation from now will contain autonomous features currently only being trialled in luxury concepts. However, a fully autonomous, driverless future? I’m not so convinced about that, for reasons which go beyond technology.