Discussed presentation schedule & various IK algorithms.

Thoughts

• CCD IK: same thoughts as previous lecture. Even though the algorithm is simple, it does make sense that there are a lot of extra considerations when putting it into practice, since it’s just iterative and not super mathematically sound.
• Was already familiar with Jacobian transpose IK from 15-462, so that all made sense. I didn’t actually know that the method was guaranteed to not increase error, so that was nice to consider.
• Inverse Jacobian methods also made sense as a more optimal solution compared to the transpose; we’re just solving an under-constrained least square problem with the pesudoinverse instead.
• Using the null space of the pseudoinverse to constrain the space of optimal solutions was also cool: just add a movement toward the rest pose constrained to the null space, so those movements necessarily don’t effect the optimal solution.