In this study, we analyse a multiple-input-multiple-output (MIMO) amplify-and-forward (AF) relay channel with inaccurate channel estimations on all links, which necessitates cooperative optimization of the source precoder, relay transceiver, and destination equalizer. Nonconvexity and the lack of closed-form solutions characterize the joint optimization problem under consideration. The optimization of a single variable in the presence of fixed others, however, has been proven to be a convex optimization problem amenable to efficient solution through interior-point techniques. In this setting, the direct link component of the AF MIMO relay channel has inspired the proposal of an iterative approach with assured convergence. For the double-hop relay case without the receive-side antenna correlations in each hop, it has also been demonstrated that the global optimality can be confirmed by solving the remaining joint power allocation using the Geometric Programming (GP) technique under a high signal-to-noise ratio (SNR) approximation, as the structures of the source precoder, the relay transceiver, and the destination equalizer all have closed forms. To guarantee the iterative strategy provides pretty decent solutions with an appropriate complexity in the latter scenario, we have evaluated their performance using simulations. When compared to a no robust system, which uses approximated channels as if they were real, simulation results confirm the robustness of the suggested architecture