This paper presents a central pattern generator (CPG)-based locomotion controller, with a foot trajectory generator and sensor-driven reflex, aiming to increase the adaptability of the hexapod robot walking on uneven terrain. The trajectory generator was used to precisely control the foot trajectories to constrain the leg movement in the workspace. The foot trajectories were further shaped by the sensor information from the proposed sensor-driven reflex, making the hexapod able to pass obstacles of different sizes on uneven terrain. In addition, a novel neck joint was designed to implement the sensor-driven reflex, with an active control mechanism for the climbing of high steps or large obstacles. Finally, a series of simulations and prototype tests were carried out to prove the feasibility of the proposed controller. The results showed that with the integration of the foot trajectory generator, sensor-driven reflex and active neck joint, the hexapod robot can achieve a stronger adaptability to uneven terrain and better performance in walking.