As described in Kinematic singularities, if there are any infinite number of solutions for a commanded pose, how can Dexter know which one to pick? The obvious answer is that it should pick the one that moves the joints the least, using the context from the prior pose to pick the new pose which is most similar.
As described in Kinematic singularities, if there are any infinite number of solutions for a commanded pose, how can Dexter know which one to pick? The obvious answer is that it should pick the one that moves the joints the least, using the context from the prior pose to pick the new pose which is most similar.
Some work on this has been done in DDE's Kin.js: https://github.com/cfry/dde/blob/master/math/Kin.js (search for
.context_inverse_kinematics
)