Just as serendipitous events in life can bring a smile, unplanned results during the taping of live robot trials can be quite interesting. In the real world robotics sensing often produces unexpected results, so why were we unprepared for the rare and actually helpful robot response during production of the above video? Humans are weird.
We chose to tape this demo in our sunroom. The location choice was based partially on size, but also for good lighting. But good lighting for video can be bad lighting for sensors. The same trial in a darker location will produce much more consistent results. We knew that.
On the day we captured this particular test procedure, we ran about 6 takes (not bad really). The previous day’s test runs were full of inconsistency because the sky was partially cloudy, and as the winds blew, resulting light fluctuations caused hot spots and shadows. But on this day, fate shone down in a more consistent manner.
That consistent sunshine influenced these strokes of pure luck:
1) The robot's final turn was probably three times as far as previous turns.
2) This fact helped the robot end its 'while' loop in the exact center of the hexagon.
Cool, but why?
At times, bright sunlight streaming in from the west overwhelmed the sensor, causing exaggerated turns on the left --or sunny-- side of the hexagon. Notice the first time around: the robot turns more than expected when on the left side of the hex. The second time around, it happened that the robot turned even farther and headed straight for middle ground.
Chance is part of life. Chance is part of robotics. Crisp examples of how chance applies in robotics are rare. For once, capturing a great illustration of sensor subtleties came as a free bonus, providing me with opportunity to coin a new word: chancenstance!
When did you last encounter interesting coding or robot development chancenstance?
Comments and questions? Checkout William's robotics blog.