"surely a cheap arse light sensor on the top will be far cheaper over the long run."
Far from it. By being able to control the lights centrally, you can make them brighter or dimmer when you choose. So, for example, the lights surrounding a football stadium can be brightened before and after the game for public safety whilst crowds are moving, and reduced to normal levels at other times.
The ability of the eye's ability to adjust to lower light levels is worth mentioning too. If you dim the lights gradually, you can considerably reduce your electricity bills without changing perceived light levels.
And a light sensor doesn't let you turn the light off completely at off-peak times, for example after 1am. Being able to avoid lighting costs for as much as half the night is a substantial energy use reduction.
Local and highways authorities are under increasing pressure not just to reduce costs but to demonstrate "green" credentials. This type of technology helps them do exactly that, particularly at a time when they're also upgrading their luminaires to LED.
On final point: bear in mind that energy costs will rise. Authorities are making cost savings now with smart street light controllers; the savings to be made in the decades-long lifespan of the controllers are highly likely to vastly out-weigh the purchase, installation and operating costs, even taking a conservative view of energy cost increases.
[Declaration of interests: I work with Telensa from time to time, although I have not been asked to comment on this article, and am writing entirely on my own behalf. Any inaccuracies are my own.]