Two performance improovments for ik and to_pin_robot
Add pre-computation for removing collision pairs in module assembly to Pinocchio robot -> 2x faster (550 µs to 350 µs with timeit).
Effect within GA loop:
After:
Before:
Refactor default ik cost function to rely less on python and more on numpy; about 33% quicker in GA optimization loop:
After:
Before:
Edited by Matthias Mayer