The paper presents Go-ICP, a globally optimal algorithm for 3D point-set registration under the $L_2$ error metric. Unlike the iterative closest point (ICP) algorithm, which is prone to local minima and relies heavily on initialization, Go-ICP uses a branch-and-bound (BnB) scheme to search the entire 3D motion space $SE(3)$. By exploiting the geometry of $SE(3)$, the authors derive novel upper and lower bounds for the registration error function. Local ICP is integrated into the BnB scheme to speed up convergence while ensuring global optimality. The method is evaluated on synthetic and real datasets, demonstrating its ability to produce reliable registration results regardless of initialization. Go-ICP is particularly useful in scenarios where an optimal solution is required or where good initialization is not available.The paper presents Go-ICP, a globally optimal algorithm for 3D point-set registration under the $L_2$ error metric. Unlike the iterative closest point (ICP) algorithm, which is prone to local minima and relies heavily on initialization, Go-ICP uses a branch-and-bound (BnB) scheme to search the entire 3D motion space $SE(3)$. By exploiting the geometry of $SE(3)$, the authors derive novel upper and lower bounds for the registration error function. Local ICP is integrated into the BnB scheme to speed up convergence while ensuring global optimality. The method is evaluated on synthetic and real datasets, demonstrating its ability to produce reliable registration results regardless of initialization. Go-ICP is particularly useful in scenarios where an optimal solution is required or where good initialization is not available.