posted on Nov, 25 2006 @ 02:16 PM
First off, I want to brag. That Chinese cat/dog looking thing is nearly exactly what I had in mind when I said (in some thread or another here) that
robots could end up ruling the battlefield given that they had the economies of a motorcycle and the agility of a race care in a dog sized or smaller
unit deployment footprint. I would still inset one pair of legs to get greater stride length and 'sitting' height even as I would probably
articulate the body for true ungulate energy recovery performance using simple springs and a universal joint. But the masted sensor/weapons pedestal
is about right, as is the ground clearance issue.
The ability to kneel the platform is equally good and while I would prefer distributed optics to a single 'head', the option of a modular,
replaceable, subassembly is nice.
Do a bit more with the aesthetics and armor angles vs. total volume enclosure as well as providing some lateral/rotational flexibility to the body or
limbs and you will be well on the way to useful quadrupedal robot that is MORE than an auto-donkey.
OTOH, the 'sentry gun' is not terribly impressive.
I don't know what the heck those low-to-ground optics are for (anti aircraft use?) but they will be masked in both surface and roofmount conditions
about 90% of the time. I would further speculate that the fixed viewing FORs are themselves faulted because you don't want to spin a heavy weapons
pedestal mount, nor even externally articulated sensor boxes on it. Both for vulnerability and mechanical reliability issues, you HAVE TO have either
a distributed array of off mount sensors feeding a single zoom-targeting aperture on-mount or a spinning mirror type 'wide angle' surveillance
system which can be _stopped_ (digitally or otherwise) to tight-field image a given magnified area.
Such is the only way to suitably refresh a full or even sectored horizon trace without being nailed to a cross of slow-to-traverse and mechanically
hard-worn reliability issues. It is also a lot harder for a threat to determine whether he/it is under observation or not if you don't bias his
awareness by turret slew (sometimes this is a good thing, sometimes not).
Imagine a Cylon Centurion and then complete the rotation rather than fix it to a raster type FOV coverage.
The tracking of the visible spectrum targets didn't impress me much as it is clear from the little green 'track surrounds' that there IS a lag
between cluttered target movement and decision-tree rejection of the fore/background imagery for the actual target 'ducking out from behind the other
side'. This tends to indicate to me that the system is _heavily_ biased towards historical motion analysis rather than true shape-definitive
'recognition'. Like tracking a lightsaber by it's glow-trail, if there is limited target area (blade is towards you or man is on-ground) it can be
deceptively vulnerable to first-to-fire initiative loss on either LAW or ATGW. The IR equivalents are much more contrast impressive but only until
you realize that these are once again non masked targets (Thellie Suit or later). I'm reminded of the images of apes moving around the encampment in
'Congo'. For a human and particularly an insurgent or SWO trying to penetrate a known-robotically defended installation to be so stupid is
Lastly of course we have the 'halt and be recognized' idiocy. I would not want ANYONE to be in front of an auto-gun without MITL intervention
during 'normal business hours' yet I swear it looks like the two volunteers are effectively being challenged and responding to a typical controlled
entry/checkpoint type scenario. Which is of course ridiculous because if you let junior get that close and he /isn't/ friendly, you are risking a
quarter/half million dollar automated pedestal to a simple suicide bomber type charge or even (again) target fixation and engagement by another
I don't know how windy it was that particular day either but if yelled voice recognition is going to work, it had damn well better be effective in
controlled conditions of ambient noise to a MUCH more distant target (100yards, minimum). The alternative again being to put one sensor on the entry
control point. And the auto-gun on a separate pedestal or mast looking down a firing lane as killsac.
The sad part of course is that by automating warfare to the extent that no man-as-soldier can occupy a battlefield without being just /butchered/ by
robotic weapons systems, you encourage a possible scenario whereby automated 'security systems' come to dominate an alternative market. One which
is often going to interface with civillians in a way that could become totalitarian without requiring a federally authorized standing army. Such
could be a real problem with corporate interests coming to totally dominate politics anyway and many of these systems starting to popup via foreign
manufacturers who don't have anything like our laws forbidding milspec weapons in the hands of industrial/corporate users (i.e. the rise of the
machines could be /very/ fast, simply because the SOA is 'imported').
At which point, Mecha will indeed become critically important. Since only they will be sufficiently sacrificial to challenge these kinds of systems
(on a magazine count basis if nothing else) and only they will then be able to enter and navigate designed-for-humans facilities and structures.