We discuss design of supervisory control logic of a multi-robot assembly cell. We present an application of the two-phase approach of Lee and Lee for supervisory control logic design. We first develop a plant model of concurrent automata for the cell. By using the automata-based model checking technique, we develop a logical controller that drives the cell so as to satisfy logical control requirements as well as carry out the assembly mission. We develop a procedure for constructing a timed global automata that models the global state transition behavior of the cell that is controlled by the logical controller. We develop algorithms for identifying conflict states and computing branching probabilities and sojourn time distributions. Finally, we model the performance controller design problem as a Markov decision process model and solve the problem by the value iteration algorithm.