Movement Platform - here's Tarzan in foreground & Conan in back

Ground clearance - Tarzan on right, Conan on left

Discussion

The choice of an aftermarket modified Traxxas Emaxx is intended to save time and effort.  It does do that, but I don't believe it's a particularly ideal solution from the performance point of view, but there just aren't any off-the-shelf vehicles available that meet even minimum requirements.  That's a separate discussion, best saved for another time and place, but if you're interested, like me, in designing and building a more ideal outdoor robot platform, or know of a new model that has become available, please contact me.

The features of the Emaxx mod are:
1.  All parts available off the web
2.  Easy assembly
3.  Spring/damped suspension
4.  High ground clearance
5.  Reliable

The problems with the Emaxx mod are:
1.  Steering is a bit wobbly and even the upgraded servo may not have enough leverage.
2.  Can't turn on it's own axis and in fact has a poor (~1.5m) turning radius.
3.  Speed range may be too high.
4.  Wheels splay and steer as the suspension bottoms out.
5.  A bit heavy.
6.  Can't climb stairs.

The usual complaint about the E-Maxx is that it's small, suspension is weak and it needs some basic aftermarket hop-ups to improve steering and speed control, which has been documented by others here, here and here and numerous forum posts all over the web.

The following is a discussion of all the chassis components in this custom E-Maxx.

Extended aluminum chassis - This will lengthen and strengthen the basic frame, which supports carrying the 10lbs or so of extra gear for the robot.   (Note - the stretched EMAXX kit appears to be no longer offered.)

Brushless motor & speed control - A brand-new product from Novak, intended for the r/c sport of rock crawling, so it should smoothly move the beast at the slower speeds typical of AGV's.  The specs for this system are lacking in detail, so it's a hope really that the speed range will be suitable.  It's rated at 119W and 2700RPM unloaded and runs well off of 2 Lithium Polymer cells.  This really isn't the right way to select a motor but I'm just hoping that it's an easy way to get going without sacrificing too much in any of the requirements.  What are the requirements you might ask?  Well, the entire drivetrain functions as a system, so here's the spec for the drivetrain: Smooth, repeatable low speed control <0.5m/s, maybe as low as 0.2m/s for the last 1/2m approaching the target objective.  Efficient at high speed >4.5m/s.  Climbing limit is largely determined by the overall weight of the craft in addition to the powertrain, so there's no spec for that atm.  No maintenance.  Reliable and durable under expected weather conditions (Northwest Winter to Texas Summer).  That's not much of a spec, I admit, but it's enough for now.  

The Novak Goat Crawler Motor and ESC gets a mixed review after using it awhile.  Quality and reliability is very high.  It never glitches or whines or does anything unpredictable.  It is silent, smooth and has plenty of torque for all the hill testing so far.  Speed testing hasn't gone beyond 1.5m/s yet and efficiency hasn't been measured,  but it "seems" very efficient.  The most I've ever had to recharge motor batteries was about 40W-hr after a couple of days of testing.  Note the motor power was always turned off immediately after each run.  People always remark about the "good looks" of the motor.  The ESC (electronic speed control) is meant for radio control of course and in my opinion is not well suited for autonomous robotics.  I'm not really sure how well the setup would work for r/c, but the main issue for Tarzan is PID-type speed control doesn't work.  When "you take your foot off the gas", the robot doesn't slow down but instead coasts.  On flat pavement, you literally have to stop the motor (PWM setting somewhere in the deadband) to brake it if you need to slow quickly.  Once this was determined, I decided to avoid getting bogged down in exploring the behavior further and for the Dallas event I used a single forward speed setting (1500 microseconds) and a single reverse speed setting (1400 microseconds), which has worked out surprisingly well.  The microsecond values refer to pulsewidths of a 50Hz pulse train that is used to control the ESC.  The PWM is generated by the 16-bit timer 1 of the atmega128 microcontroller and puts no load on the cpu in so doing.  Timer 1 is set in "phase and frequency correct PWM mode" and it provides both the 50Hz timebase and the precise pulsewidth (to 4 microseconds).  This arrangement has been ultrareliable but the coasting behavior makes speed control a bit tricky.  By the way, the actual cause of the coasting may not be due to the motor so much as the EMAXX drivetrain.  I just don't understand the mechanics well enough at this point to be clear on that.

Wheels and Tires - 7" diameter is the key here, compared with stock 5.75" wheels.  Already mounted tires with foam inserts is a nice time saver.  The only issue with these wheels is they are made for a 17mm hex mounting system, while the E-Maxx standard is 14mm.  Just make sure they come with adaptors, which mine did from Tower Hobbies.  There are several other mounting standards for various monster trucks so be careful.

Suspension - The stock suspension is too weak for a heavy robot, so it is mandatory to at least increase the spring size.  I've opted to replace all the components of the suspension as follows:  Titanium arms from flextekrc.com, big-bore shocks from Traxxas, heavy-duty springs from Trinity, aluminum axle carriers (knuckles)  from Tower, Fastlane aluminum bulkheads and shock towers from rc-monster.com, but so far the axles and remaining components are stock.  

Bearings - A complete set of ceramic (silicon nitride) bearings for the E-Maxx was obtained from Tower.  The purpose is to reduce rolling friction.  The claim of 30% reduction remains to be validated but the point is to reduce power requirements as much as possible.  

Steering servo - Upgrading the servo torque is needed for a robot version of E-Maxx.  We'll see if my choice is adequate for Tarzan.  I am opting to go without a servo saver to begin with in order to maximize steering precision.  This is a debatable choice.  We'll see.

Transmission - An aluminum case from Fastlane and a one-motor adaptor from GorillaMaxx were ordered from rc-monster.com.  A hardened steel idler gear, 72 tooth spur gear and 12 tooth pinion gear were ordered as well.  The gearing ratio will be 47.3:1 in 1st and 29.37 in 2nd, relative to wheel revs.  The standard gearing is an 18 tooth pinion with a 66 tooth spur, which produces ratios of 28.91 and 17.95 respectively.  Factoring the motor, wheels and gearing together, I'm expecting a top speed of 2700/4800 * 17.95/29.37 * 7/5.75 * 40mph = 0.56 * 0.61 * 1.2 * 40mph = 16.4mph.  If that turns out to be too fast, then I can add a 2:1 reduction unit on the motor and change the pinion/spur ratio a bit in order to hit about 10mph top speed.

There are probably other non-stock components on Tarzan that aren't listed above.  Sorry bout that, but they're not critical and it isn't difficult to research all the options online and do it your own way.