FIGURE 4. Flipping the board around reveals the magnet
and 12-bit 360 degree AS5045 magnetic potentiometer.
FIGURE 6. Motor test setup using the hexapod body, half of a femur, tape,
and a solid metal weight.
60
50
40
30
20
10
FIGURE 5. The gearbox of the
MX- 28, very similar to the RX- 28.
2. 71 2. 5
Time (s)
Time (s)
FIGURE 7. P and PD comparison. The pure P controller (red) overshoots the goal
position of 50 degrees, but the implementation of the derivative feedback term in
the PD controller (blue) reduces the overshoot. Both controllers still exhibit an
error from the lack of an integral term.
Control
System
LINKS
My Personal Website
As previously discussed, though
some control systems may work
well for different scenarios, no one control system can work
in all situations. The control system used in the RX- 28
involved setting a slope and certain limits. Though easy to
understand, the controller was non-linear and difficult to
adjust for most robotics projects. Under the right tuning, the
motor could be of good use for human-robot interaction,
where it is desirable to have the motor move quickly to a
position but still remain compliant when getting physically
moved in unpredictable ways.
The PID controller is much more useful when operating
many systems, especially those that can be modeled as a
mass-spring-damper system. Figure 6 shows a simple setup
with the motor strongly clamped to a table, with half of a
hexapod femur and a heavy iron weight taped to the end.
SERVO 01.2012 59
