As of now, the fixed initial position that was used for the optimization was a position I found previously by evaluating a number of trajectories and choosing a position that seemed to end up in a more periodic trajectory.
However, perhaps this parameter could be better initialized dynamically by solving a smaller optimization problem for finding a near periodic/somewhat stable trajectory and then using that solution's initial position?
As of now, the fixed initial position that was used for the optimization was a position I found previously by evaluating a number of trajectories and choosing a position that seemed to end up in a more periodic trajectory.
However, perhaps this parameter could be better initialized dynamically by solving a smaller optimization problem for finding a near periodic/somewhat stable trajectory and then using that solution's initial position?
_Originally posted by @rasmusmarak in https://github.com/rasmusmarak/TOSS/pull/46#discussion_r1283149422_