TiggerBot
This was my first robot, built in January 2002. Its major failings were a very
low torque to weight ratio and exceptionally inefficient power management.
2007-11-01
After a few years more experience, here is some self-criticism:
- H-bridge switch selection. You must be very careful with darlingtons.
They are dirt cheap and have high gain, so as a robotics neophyte they seemed
appealing. However, the forward voltage losses meant that as much power goes
into the transistors as the motor.
- PIC processor selection. I got a programmer and a bunch of these chips
for free at one point. There was no technical reason for choosing them. PIC
vs AVR may be a religious issue, but my faith is entirely with AVR at this
point.
- Assembly code. Again, it's what I had. Also, there is really no such
thing as a C compiler for low-end PICs, (they don't really have stacks) though
there are programs that do a good job of pretending.
- Radio modem selection. I used a half-duplex 433 mhz on-off keyed modem
pair. They were subject to interferrence and only good for like 4800 bps. I
had to implement a communication protocol with checksums and retries.
Strangely I recall that I paid more for them than for the maxstream units.
- Tread geometry selection. This only became apparent when I got into
working on TiggerBot II. Tread-laying vehicles without suspensions are prone
to doing backflips instead of climbing over obstacles and traction on uneven
surfaces is limitted at best.
- Battery configuration. Tiggerbot had two separate 4.8v NiMH packs -- one
for the MCU and sensors and one for the motors. A lot of hobbyists think that
having separate batteries for logic and motors is a good idea. They are
completely and categorically wrong.
original page:
hardware specs:
- tamiya bulldozer chassis and drivetrain
- PIC16C74A on-board microcontroller
- TIP122/TIP127 h-bridge reversible motor control with 10 bit pwm
- optical tread encoders with 8 tick/inch resolution
- front and rear bump sensors
- forward looking turret mount Sharp GP2D12 infra-red rangefinder
- forward looking turret mount Devantech SRF04 ultrasonic rangefinder
- 433 MHz 2400 baud radio modem
- two 4xAA NiMH battery packs
- weight, including batteries: 0.66 kg (~1.5 lbs)
- 24 cm long, 11 cm wide, 15 cm high (minus antennas)
- black, orange, and of dubious intellect - like tigger.
software specs:
- firmware: runs in the pic16c74a. controls programmable movement and collects sensor data. communicates using a simple packet format to alleviate the dreadful error rate the radio modems have. tigger-0.010.asm - status: currently ~2000 lines, mostly done.
- software: runs on the controlling computer. does packet communications, conversion between disgusting robot command sequences and clean C++ api, object mapping, X interface, interactive driving, etc... status: after much hacking, the low level stuff is pretty slick. currently 4100 lines of C++ total, nowhere near complete. (complete? what's that?)
