projectStateToObstacle : now take optional argument rotation to define a desired rotation of the end effector. If this rotation is provided, the normal argument is ignored (todo : at least check that the rotation and the normal values are consistent, and maybe always rotate to match the normal if not).
when usePosturalTaskContactCreation = true correctly allow the rotation to move along the contact normal, and not along the z axis.
contact generation :
if usePosturalTaskContactCreation = true and the projection fail, try again without
fixedStep heuristic : if called without information about the future root's path, use the static heuristic.
projection : closestPointInTriangle remove the hardcoded epsilon used to move in the direction of (orthogonalProjection -> projection inside the triangle). This change have been made because when the orthogonal projection is already inside the triangle, the numerical approximations may lead to a direction which go inside the obstacle.
Bugs fix :
rbprm shooter : fix error in computation of the weights for each triangles
RbprmNode::chooseBestContactSurface fix a segfault where bestReport could be unitialized when the computation of the intersection fail. (FIXME : I still don't understand why computeIntersectionSurface can fail)
changes in projection :
rotationto define a desired rotation of the end effector. If this rotation is provided, thenormalargument is ignored (todo : at least check that therotationand thenormalvalues are consistent, and maybe always rotate to match the normal if not).usePosturalTaskContactCreation = truecorrectly allow the rotation to move along the contact normal, and not along the z axis.contact generation :
usePosturalTaskContactCreation = trueand the projection fail, try again withoutclosestPointInTriangleremove the hardcodedepsilonused to move in the direction of (orthogonalProjection -> projection inside the triangle). This change have been made because when the orthogonal projection is already inside the triangle, the numerical approximations may lead to a direction which go inside the obstacle.Bugs fix :
RbprmNode::chooseBestContactSurfacefix a segfault wherebestReportcould be unitialized when the computation of the intersection fail. (FIXME : I still don't understand whycomputeIntersectionSurfacecan fail)