1: \begin{abstract}

2: This paper proposes a new method for rigid body pose estimation based on spectrahedral representations of the tautological orbitopes of $SE(2)$ and $SE(3)$.  The approach can use dense point cloud data from stereo vision or an RGB-D sensor (such as the Microsoft Kinect), as well as visual appearance data as input. The method is a convex relaxation of the classical pose estimation problem, and is based on explicit linear matrix inequality (LMI) representations for the convex hulls of $SE(2)$ and $SE(3)$. Given these representations, the relaxed pose estimation problem can be framed as a robust least squares problem with the optimization variable constrained to these convex sets.  Although this formulation is a \emph{relaxation} of the original problem, numerical experiments indicates that it is indeed exact -- i.e. its solution is a member of $SE(2)$ or $SE(3)$ -- in many interesting settings.  We additionally show that this method is \emph{guaranteed} to be exact for a large class of pose estimation problems.

3: 

4: %This convex relaxation formulation can take advantage of modern semi-definite programming

5: %algorithms, and also allows for an $l_1$-regularization term to minimize outlier effects.

6: %Advantages of this approach include: 1) lack of local minima in the estimation problem (in contrast

7: %to ICP methods); 2) second order convergence of the algorithm; 3) minimization of outlier

8: %corruption, which are common in stereo vision; and 4) ease of implementation via modern semidefinite

9: %solvers. The approach is illustrated and compared to other methods using a pose estimation problem.

10: \end{abstract}

11: