Feedback Results
Comments:
- To recognize a chair, it can use some other distinguishable properties,
instead of color.
- Didn’t see wandering. However, the idea behind random “goals”
seemed very functional.
- Use of ACTS to detect the difference in chair/obstacle worked efficiently.
- Chair didn’t seem to end up where it should at the desk. Some improvement
seems needed to place the chair somehow and not the robot.
- Watch out for the laser’s status lights or change your color to something
not normally visible.
Which aspects of the RAFS project particularly impressed you? Please explain.
- Smooth navigation – It’s fast to detect obstacles in the closed
world.
- Grabbing a chair – It has a pretty strong grabber and performed well.
- Path planning and movement of chair to the desk. To see it when fully functional
would be quite interesting.
- Path finding, chair recognition – [fair].
- Chair gripping seems like one of the hardest modules to implement. The
robot’s reorientation was impressive.
- The complex control of the robot from a PC.
- I saw you are using color recognition to detect chair, don’t you
think coupling that with edge detection would help to find a chair or desk.
- ACTS recognition seemed to work well.
Which aspects of the RAFS project did you feel needed significant improvement?
Please explain.
- Objects Orientation Recognition – It should figure out the shape
of an object.
- Localization Precision – It should place a chair in the right direction.
- Again, just the movement of the chair. Also the integration of movement
with the chair and forward sensing, seems objects dead on could be easily
hit instead of avoided.
- It had some problems (getting stuck) before reaching the chair. This could
be improved, though it was working previously.
- Code just needs a little more work.
- You might want to rethink the Placement Algorithm.
|
|
|