 |
|
|
Beginning in the 2020s, development of
autonomous robots began with a fury. Battlefield requirements,
commercial endeavors, and first-responder needs all had problems that
these AI filled robots could solve. The first breakthrough was a device
called the standardized robotic coding appliance or SROCA.
Developed by the father of modern hybotics, William Pratt, SROCA allowed
any of the locomotion, defensive measures, sensory, or self-preservation
directives of a robot to be controlled from a commonly available module
that was standardized across all AI controlled robots. Pratt designated
robots by AI function, such as military, mining, search & rescue,
construction, etc. The locomotion platform, such, as wheeled or
tracked, didn’t matter since the SROCA could adjust for either. For
example, a wheeled military robot could be converted to a tracked robot
during the mission, without the need to be shut down or taken out
of service. In fact, such a conversion could be accomplished in the
field by two support personnel in 4 minutes.
The main drawbacks to AI robots were the
large amount of coding required, as well as the changes to the
operations format. To use our military robot example again, a robot
guarding a particular entrance to a secure complex required over 30
hours of downtime to be turned into a robot guarding a building inside
of the secure complex. In 2045, that all changed.
Dr. Bernard Adrian developed Hybots.
Hybots were the same as any other robot but with one particular
and revolutionary difference: the brain wasn’t a chip, it wasn’t organic – it was both.
It was neurons cultured from human cortex, formed into a network that not
only reacted to external stimuli, but learned from it. Instead of
complex AI robots running massively complex software routines for every
possible condition, Hybots were given simple instruction sets and
expected to learn their way to a successful mission. Hybot brains
were expected to last about 24 to 36 months – long enough to complete
even the most complex mission it was assigned.
Dr. Adrian discovered that several
interconnected neural networks could be placed into one Hybot brain
allowing for amazing results. The processing of information by these
interconnected neural networks is done in parallel rather than
sequentially so each neural network can be program with a unique set of
general orders. Since the Hybot relies on these interconnected neural
networks collectively to perform its function, a unique property of a
neural network is that it can still perform its overall function even if
some of these interconnected neural networks are not functioning: they
are very fault tolerant.
The technique to program the Hybot’s
neural networks with a set of industry specific general orders was
called general order stuffing or GOST setup.
Mission specific orders, called
specific operational order kit or SPOOK setup, added
the details. Using our military robot example again, all of the Hybots
receive the same military GOST setup; the Hybot would just have to be
told, either verbally using password protected commands or via an
encrypted radio, what to guard, for how long, and what force was
pre-approved. To change from guarding the front gate to a building
inside would take 2 minutes.
The Hybots were engineered
to last only 24 to 36 months but it appeared that they had the ability
to repair failing neural interconnects and to remember SPOOK setups
indefinitely, even when they were replaced with other SPOOK setups.
Hybots had a lifespan of over 200 years and could perform self
maintenance whenever needed. The Hybots even learned to disable the
standard and emergency shut down controls that were part of the GOST
setup. All of these things were discovered however, after they had
been deployed to Mars.
In 2115, about 200 hunter-killer Hybots were developed and sent to Mars
in an effort to rid the planet of the infestation of uncontrolled Hybots.
These hunter-killer Hybots were loaded with special SPOOK setups:
destroy with extreme prejudice any wheeled, tracked or footed entity
that didn’t have the correct transponder code response. Each
hunter-killer Hybot was fitted with an MPC and 500 rounds of ammunition
(MPC-P-S-FS: Slug-type, tungsten ferrosodium, fin stabilized). They
only needed approximately 5 kills each to rid Mars of the menace: good odds the
engineers thought, since it would only take about 1 to 2 rounds to
disable the unruly Hybots.
As a failsafe, the hunter-killer Hybots had an external explosive charge
that was set to auto-destruct after 36 months. Unfortunately, all of
the hunter-killer Hybots stopped communications with the control
spacecraft within 3 months after landing on Mars. The engineers
assumed the auto-destruct mechanism functioned as planned.
Silly fools.
-
David C. Gann, Senior Archivist / Cook |