gtsam

Author	SHA1	Message	Date
Frank Dellaert	7d4de3ec20	Added a 1D manifold implementation of 2D rotations. The new representation stores (cos theta, sin theta) rather than theta itself, ensuring that (a) rotate and unrotate do not call cos/sin, (b) same for all derivatives of rotate and unrotate, (c) when you call angle(), you always get the standardized answer given by atan2. The idea is that we will use this new type in Pose2 instead of angle.	2009-12-09 21:50:27 +00:00