We attached an inverted pendulum (constrained to one degree of freedom) to the Evolution platform. We succeeded in tracking the angle of the platform and developed a sophisticated model that would allow the robot to balance the pendulum. Unfortunately, when these systems were integrated, it was discovered that the robot was only able to make approximately two updates per second—not fast enough for it to keep the pendulum upright. We believe that our approach is sound; having a faster camera and faster access to the motors would have resulted in our success. Success might also have been achieved with a pendulum with more weight; however, we were not able to test this.
The remainder of this report is divided into the following sections:
Presented here in chronological order
|Behold, the inanimate
|The hinge we attached it to.
The hinge was attached to the frame by velcro, but ended up duct taped, as well.
|The testing setup|
|The camera's view of the test setup|
|And the result.
Note that most of the slopes were close, while a few were way off. Also, our up and down are reversed, so the slope should be negative.
|The camera's view of the final system. Note the green background contrasting with the red pendulum.|
|A few mis-typed constants into some matrices resulted in some
rather excessive accelerations. The units are in cm/s.
Had the robot been able to sustain this acceleration, it would have achieved light speed in approximately 1/6 of a second.
Surprisingly enough, this did not occur.
|A simulation of how our system would respond given a sampling
rate of 2 Hz (which is approximately what it had).
Note that the units for theta are radians; this means that the pendulum has rotated nearly completely around.
|Our final robot|