Foot drop is the phenomenon that ankle joint cannot produce dorsiflexion and the toe lifting is incomplete or impossible due to nerve control dysfunction. The functional electrical stimulation (FES) is used as a treatment method to correct foot drop gait, which uses low-frequency pulse to stimulate tibialis anterior to cause muscle contraction and dorsiflexion of ankle joint. The FES output intensity modulation method based on EMG modulation and iterative learning control (ILC) is proposed in this article. The angular velocity signal of the lower leg is used to predict the EMG of tibialis anterior in healthy gait through the dynamic BP neural network, and the toe pitch is used as feedback signal to output the reference EMG through ILC. The reference EMG and the EMG predicted by neural network are weighted-average to obtain the modified EMG. Finally, the FES output is modulated by the muscle activation characteristics. The experimental results show that the toe pitch angle in open-loop EMG modulation mode is only about 17°. Through the closed-loop modulation mode, the maximum toe pitch angle is about 21°. By analyzing experimental data. It can be concluded that the system can help patients with foot drop to carry out rehabilitation training.