To solve the online path planning of multi-robots in dynamic environments, a novel secondary immune responses-based immune path planning algorithm (SIRIPPA) is presented. The algorithm comprises two immune stages. In the primary immune stage, the antibodies are mainly designed for obstacle avoidance and a primary immune kinetic model is designed in terms of the different impacts of obstacles on robot behaviors. The primary immune antibodies and their concentration values are mainly taken as the prior knowledge to accelerate the secondary immune response. In the secondary immune stage, aiming at the same obstacle antigens, which invade once more, the immune system quickly produces many behavior antibodies. Combining the primary immune results and secondary immune response results, the path planning performance of multi-robots is improved. The simulation experiment indicates that, in static environment tests, compared to corresponding immune planning algorithms, the SIRIPPA exhibits an average reduction of 6.22% in the global path length, a decrease of 23.00% in the average smoothness, and an average energy consumption reduction of 27.55%; the algorithm exhibits a better performance for path planning. The simulation test in a dynamic environment shows the good flexibility and stability of the SIRIPPA. Additionally, the experimental results in a real environment further support the validity of the SIRIPPA.