search

nepfaff / tenaci

Software for a robotic manipulator that can rotate and stack cubes, draw sketches, and prepare a miniature breakfast.
0 stars 0 forks source link

Specific waypoints for task 2 video #13

Closed nepfaff closed 2 years ago

nepfaff commented 2 years ago

The general task 2 algorithm will have some error due to different effects of gravity at different distances from the arm base.

We would score better on the video part if we only use the general algorithm for demo day but slightly adjust the waypoints manually for the video.

This error is due to theory meeting the real world. Hence, this manual adjustment must be done with the real robot based on trial and error. The best way to do this is to manually extract the waypoints that the automatic algorithm would produce. List these so that they can be manually modified. Then manually modify by adjusting values or adding additional waypoints.

TODO Before Lab: Extract list of waypoints that the algorithm would produce. Do this for tasks 2 a, b, and c. Create extra config functions for this, e.g. getWaypointsForVideoTask2a. TODO: During Lab: Finetune the waypoints using trial and error. Maybe ensure that we get perfect results 3 times in a row if we have time for this. Could already do a bit of this before the lab based on the videos recorded during the last lab.

nepfaff commented 2 years ago

These have now been extracted into .mat files.

Modify these by:

  1. Load them by clicking on the .mat file (Using the complete main waypoints results in the start pose being included which must be manually deleted before saving)
  2. Open the loaded struct list by clicking on the waypoints environmental variable
  3. Edit values or add/delete rows
  4. Save into the same file using ctrl+s
nepfaff commented 2 years ago

done and recorded