Gait asymmetry in stroke patients severely impairs their mobility and exoskeleton-assisted rehabilitation therapy is an efficient solution for gait recovery. However, existing research primarily focuses on improving gait symmetry, with little consideration given to the potential risks associated with the human-robot interaction (HRI) force. To address this issue, this paper proposes a dynamic planning method for symmetric gait trajectories in exoskeleton-assisted rehabilitation based on the HRI force. This method dynamically generates symmetric gait trajectories to improve gait symmetry while keeping the HRI force within a tolerable range. The HRI force is estimated using an enhanced unknown system dynamics estimator (EUSDE) and the gait information is extracted through an adaptive oscillator. The EUSDE-based tracking controller is developed for the impaired side in order to enhance gait symmetry. Finally, simulations validate the effectiveness of the proposed method.