Cost

Our primary goal is to decrease the cost to about $500. We believe this can be done with bulk purchases of the materials. Consolidation vendors, anyone?

RoboStix

GumStix is coming out with the RoboStix microcontroller, which plugs into the Gumstix, replacing the I/O board. The RoboStix is spec'd with an Atmel CPU, digital ins and outs, servo outs, analog ins, serial, and regulated power. If the price is right and the board turns out to be easily codable, we may replace the Brainstem with the RoboStix. Certainly this will make the electronics significantly smaller on top of the robot.

Motors

Probably the single biggest weakness in the current design is the use of continuous-modified servos for the wheels rather than encoded motors. This was largely a cost decision: motors require motor controllers and H-bridges, etc., and servos are ubiquitous. But servos were not designed for this task and will likely burn out or strip their gears eventually. But what motors and gearboxes to switch to?

Aluminum

Good precision-designed robots are often made out of machined aluminum. We went with PVC because it was easier for us to handle without special tools (as should be painfully obvious, we're not mechanical engineers). PVC also doesn't conduct electricity. But in the future, a single aluminum base might be attractive.

Batteries and Chargers

The current batteries last a surprisingly long time; but a larger battery can be fitted in the robot. We think a sub-C cell hump pack will probably fit well. Coming up with an auto-charger for the robot would also be an interesting challenge.

Forward Casters

The robot occasionally tilt forwards a bit, then flops back. We knew this when designing the bot, and we put as much weight as we could right on the caster to compensate. But putting casters in front is not mechanically feasible: when going over a bump, the wheels would be lifted off the ground. However some anti-tilt bars or something might be called for in the front at some point.