EVER noticed how road-sweeping trucks just seem to rearrange dirt and garbage
rather than removing it? Make the trucks smarter and they will do a lot better.
You just have to arm the vehicles with a laser, a digital camera and an
image-processing computer.
The problem with standard road sweepers, says Graham Parker at the University
of Surrey, is that they鈥檙e dumb. The settings for the brushes鈥攖he speed
and the force with which they鈥檙e pushed onto the road, for example鈥攁re
usually set at the beginning of a run and then left alone.
But some rubbish needs only light sweeping while other types of debris, like
patches of spilt builder鈥檚 gravel or wet sand, call for more vigorous scraping,
so having just one setting makes the sweeping process inefficient. 鈥淭hese
vehicles are often not that effective鈥攕weeping becomes a two-man operation
with one in the truck and another behind on foot sweeping up what the truck
misses,鈥� Parker says. Identifying the types of trash to be dealt with should
help the truck to pick up more rubbish at the first attempt. This will remove
the need for a second pass, speeding up the operation and cutting fuel
consumption and diesel emissions.
Advertisement
So with his colleague Gareth Peel, Parker set about making the trucks
smarter. Next week they鈥檒l tell a conference on Advanced Intelligent
Mechatronics in Como, Italy, how they plan to do it. A laser mounted near the
front projects a line onto the road ahead. A digital camera in the truck focuses
on this laser line and sends images to an on-board computer.
Parker says that an image processing program monitoring the shape of the
laser line can pick out and identify lumps of rubbish lying in the road. This
information could then be used to move the truck鈥檚 brushes to intercept the
litter, or to raise the alarm if the debris is in danger of clogging up the
vacuum unit.
For some types of litter, though, the system is smarter still. It detects
patches of gravel, sand or leaves on the road because they tend to have
irregular outlines. These are then distinguished by measuring variations in
their colour and brightness. 鈥淚f you have a few dry leaves, you don鈥檛 need
anywhere near as much power as with wet sand,鈥� explains Parker.
鈥淚t鈥檚 great to see people working on practical problems,鈥� says Mark Atherton
of the mechatronics group at South Bank University in London. But he adds that
the system would need a manual override in some situations. 鈥淚f it came across a
patch of concrete stuck to the floor, you could be there a while.鈥�
