Microcontrollers (Robot Brains)

I've worked with a bunch of different microcontrollers - some of which have made their way into combat robots. All of them seem to have their pros and cons. These chips fall under two different primary categories.

Interpreter Chips

They typically have the following built into a single "chip":
CPU
ROM (for the interpreter and program storage)
RAM (for variables)
An Oscillator (to control clock-speed)
Voltage Regulator

A "compiler" on your PC converts your BASIC code into "P-CODE" and sends it to the chip via a serial connection. A program on the chip called a bootloader then writes the P-CODE to an area in ROM.

P-CODE is a kind of intermediary instruction set between a high-level language like basic and a low-level language like assembly.

The ROM comes preloaded with an "interpreter" program that (slowly) converts P-CODE into machine language as it's executed.

The main upside to this setup is that since you're not really "burning" the chip - you don't need a separate development board and programmer.

Since this setup is generally geared towards hobbyists - the development kits tend to be reasonably priced and well documented.

It also has its draw-backs. Since your code is interpreted real-time instead of compiled down to assembly language - there is a -major- performance hit. However, from my experience most tasks just aren't CPU intensive enough for raw computational power to be a major issue.

The interpreter also significantly cuts into program space and variable memory.

Non-Interpreter Chips (I'm sure there's a better term for this)

The good news is that you don't have to hand-code the machine language or even necessarily use assembler. For most chips there are a variety of compilers in high-level languages like C or BASIC.

For more information see the review of the Microchip PIC16F88