The experiment, tracking a continuous path(e.g. a circle), is often used to check whether an IK method is useful or not.
When you track a given path, you need to solve IK for every path point, and it is request that the joint angle should not jump greatly, which means we want to obtain smooth joint angle curves.
But if we use GGIK (a sampling method), it must jump greatly between a pair of adjacent path points. Other methods such as tracik, may use the IK solution of the former path point as the initial_q to solve for IK of the later path point.
So I want to know, is there any method (such as take specific samples when eval_forward, or some other methods) for GGIK, to minimize the joint jump and execute the path tracking task?
The experiment, tracking a continuous path(e.g. a circle), is often used to check whether an IK method is useful or not.
When you track a given path, you need to solve IK for every path point, and it is request that the joint angle should not jump greatly, which means we want to obtain smooth joint angle curves.
But if we use GGIK (a sampling method), it must jump greatly between a pair of adjacent path points. Other methods such as tracik, may use the IK solution of the former path point as the initial_q to solve for IK of the later path point.
So I want to know, is there any method (such as take specific samples when eval_forward, or some other methods) for GGIK, to minimize the joint jump and execute the path tracking task?