|the skeptics continues to
match or exceed that expectation.
A Closer Look
We now look closer into land based, servant class
mobile robots. The typical PC power consumption is generally only
a couple hundred watts at 110VAC and when connected into our home
or office power grid, we think little of it. This same PC powered
by a heavy, fifty to sixty pound battery capable of eighty ampere-hours,
only lasts two to two and a half hours.
The intrinsic high cost of brains on board is not the
dollars expended for the several-hundred megahertz CPU's, but the
severe battery life degradation due to their enormous power appetite.
The gigahertz systems now available are easily fifty percent more
power hungry. Since battery life and system efficiency are significant
cost drivers in laptop computer and Personal Digital Assistants (PDA)
design, it is unlikely that a laptop and/or PDA will have equivalent
computational capabilities to that of a desktop PC, at the same price
point, anytime soon. As the U.S. space program clearly demonstrated,
reducing size and increasing efficiency carries significant cost premiums.
In summary, a traditional gate for truly utilitarian capable
personal robots has been battery life - a direct result of needing
extensive computational power. For a mobile personal robot to have
any true utility, it must be capable of sustained activity for more
than one to two hours. Its run/recharge ratio must be more human friendly.
i.e. we sleep (recharge) six to
||eight hours, and run (work)
sixteen to eighteen hours routinely. For a personal robot to fit in
humankind's world, running only one or two hours and then requiring
recharging for four to six, places them in the same rich man's toy
category as the first horseless carriages that had limited operating
time (range) and difficult operations.
Based on the discussion so far, it seems that "brains
off board", connected through an RF data communications link
reduces power requirements. While this observation is quite correct,
we need to examine additional issues unique to mobile robots. Fortunately,
personal robots have benefited from a confluence of relevant technologies
now available that were not cost effective fifteen years ago.
Industrial robots tend to reside in very stable and
repeatable environments, often in blocked off areas of operation.
In contrast, mobile robots, and personal robots in particular, live
in a dynamic world requiring many different sensor systems to detect
changes in their environment and behave appropriately. Since industrial
robots stay in manufacturing work cells, the number of real time systems
requiring constant supervision is relatively low. A mobile robot,
due to its need for multiple, complementary sensor systems, requires
a great deal of real time management and consequently many Microprocessor
Control Units (MCU's).
By moving of the computational power off board to save
power and cost, there is now a requirement for MCU's. Now the CPU
is no longer