Focusing on hybrid flow shop robotic cells scheduling problem with blocking, for optimizing simultaneously robotic operation sequence and part input sequence, a branch and bound method was proposed. Firstly, robotic activity was defined to transfer double sequence into single sequence. Secondly, order insertion rule was proposed for obtaining feasible solution. Finally, according to the order insertion rule, branching process was designed. Stochastically generated instances were computed by CPLEX and branch and bound method. When there were three tanks in robotic cells, the value of objective function got by branch and bound method was the same as that of CPLEX; but compared to CPLEX, the average computation time was reduced by 38.58%. So the branch and bound method was effective. When the number of tanks exceeded three and the computation time did not exceed 600 seconds, compared to CPLEX, 85.19% of instances could be found out better solution. It means that the branch and bound method is more valuable, especially for big scale problems.