In order to enhance the autonomous navigation ability of multiple mobile robots, a tabu search based leader follower flocking algorithm is proposed to overcome the local minima problem and realize the autonomous navigation in unknown obstacle environments. Leader robot is capable of avoiding obstacles and seeking goal by tracking virtual sub goals generated by tabu search algorithm, follower robots which are led by leader robot keep formation and cohesion under the control of flocking algorithm for motion control. The whole flocking system is able to go through obstacle environments and reach the preset goal without collision and separation. A simulation example in complex environment validates the correctness and effectiveness of the proposed algorithm.