From constructing my MazeBot, I learned it can be quite difficult to accurate control the movement of a differential drive robot with normal motors, especial the low-voltage (high current) one from the Tamiya twin gearbox. This brought up the question if stepper motors would perform better.
Mechanics
This robot is a differential drive robot, using two stepper motors and a Tamiya ball-caster at the front. The wheels are from BalBots.com and Aloys kindly made adapters to fit the (non-metric) wheel hubs on the motors.
Electronics
The
robot is controlled by a Microchip Pic 16F877A. One ULN2806 is used to
drive both motor, using half-stepping. Half stepping almost increases the
power consumption by 50%, but also makes the motors more powerful.
The line-sensor consists of 8 Fairchild QRB1134 IR Photo reflectors. The line position is determined every 2ms by reading the 8 photo-transistors with and without the IR-led on. The difference is an indication of the reflection and the total reading of the sensor is calculated by multiplying the difference with the sensor number, add all these values up and divide the sum of the differences.
The power supply consists of 12 NiMh penlite batteries and make the robot run over 30 minutes.
Results
The main lesson learned
from this project is that stepper motors are well-suited for low
rmp's without the need of gears. And they give
- as expected - predictable results without a control loop. Stepping
both motors at the same speed makes the robot drive in a - really close
to - straight line.
This comes however at a price: the power consumption of the stepper motors is high and the torque is low, compared to normal motors of the same size or power consumption. This torque decreases even more when the speed increases.
I ran the robot at 10 cm/s, which is close to the maximum for this robot. When it encounters an obstacle, the motors are stalled and have just enough torque to restart when the obstacle is removed.
It is possible to run the motors at a slightly higher speed, but this requires gradual increase of the speed and feedback from the wheels to verify if the wheels are turning as intended. In this case it might be better to use normal motors though.
The robot competed in line following at the 2006 RobotMC open club championships. It completed the appx 5.5 meter course in 56 seconds, earing the second place.
After the 2006 championship, the robot is modified to compete in the line maze competition.
Links:
My Robots