Virtana Internship Week 4
This week at Virtana has been spent attempting to build upon my work from last week. I have updated my calibration target code, added the initial work for a new scene to the simulation, and made a tool to do a perspective warp for images.
My calibration target was a bit off so it had to be updated. I had to go over the mathematics for coordinate vertices and texture vertices.
I have been working on adding a new scene to the simulation. At first, I was creating collision and visual shapes, and then making individual pybullet multibodies for each part of an object. This meant that after I created the multibodies I had to join then by constraints. I had successfully joined the tables but this apparently wasn’t the best way to do things. I didn’t realise that the function to create multibodies had parameters for links until my boss pointed it out to me. This meant that I could create the whole object as a single multibody and not have to fiddle around with constraints. This gave better results for my objects.
The last thing I did this week was create a tool to do a perspective warp on images. I originally had tried to determine the bounds of the images and do the perspective warp on that using OpenCV methods but this didn’t work out. I moved on to just determining the points myself and just passing that the OpenCV perspective transform method.
Once I made finished the tool, I had to look at a bunch of images and record the points of the main focus of the image to feed into my tool.