This paper presents a cooperative estimation of the pose of an object in multi agent systems, e.g., camera networks. The individual measurement of the object pose is assumed to be Gaussian (on SE(3)). We investigate an iterative scheme for the agents to cooperatively compute the optimal pose of the object, which minimizes a maximum likelihood cost, in a distributed fashion. By linearizing the pose perturbations at the optimal pose we formulate the problem as a linear maximum likelihood (ML) estimation at each iteration. Then, by utilizing the average consensus protocol, we show that the global ML solution at each iteration can be estimated by the agents with linear convergence rate, under the assumptions that the underlying graph of the system is connected and the weight matrix is double-stochastic.