This paper proposes a mixed neural network (NN) for a hexapod robot locomotion control. The mixed NN consists of a seven-node fully connected recurrent NN (FCRNN) gait controller for straightforward walking and a feedforward NN (FNN) controller for executing wall-following behavior. The FCRNN controller controls each hip joint of the robot to generate a walking gait with the objective of improving walking performance. The FNN wall-following controller is characterized by a small network size to realize the function of detecting an obstacle and walking along its boundary. The two controllers are learned through a data-driven multiobjective evolutionary algorithm. A training environment and multiobjective functions are designed to perform evolutionary parameter learning of the two controllers using the non-dominated sorting genetic algorithm-II (NSGA-II). Simulation results show that the walking gait generated by the FCRNN gait controller is faster than that generated by the typical four-state finite-state machine. Simulation results in training and test environments also show the effectiveness of the simple FNN in the mixed NN controller for executing the walking following behavior.