Purpose -This paper aims to construct a central pattern generator (CPG) network that comprises coupled nonlinear oscillators to implement diversified locomotion gaits of robot AmphiHex-I. With the gaits, AmphiHex-I will have a strong locomotion ability in an amphibious environment, which is motivated by a novel public health application to detect the amphibious snail, Oncomelania hupensis, the snail intermediate host of Schistosoma japonicum, as an amphibious robot-based tool for schistosomiasis surveillance and response in the future. Design/methodology/approach -First, the basis neural network was built by adopting six Hopf nonlinear oscillators which corresponded to six legs. Then, the correlation between the self-excited harmonic output signals generated from CPGs and various gaits was established. In view of requirements on its field application, the authors added a telecontrol system and an on-board battery to support the real-life remote control and a high-definition camera and a global positioning system module to acquire images and position information. Finally, the authors conducted the testing experiments on several tasks, e.g. detecting the distribution of Oncomelania hupensis snails. Findings -The results demonstrate that the CPG is effective in controlling the robot's diversified locomotion gaits. In addition, the robot is capable of fulfilling several testing tasks in the experiments. Originality/value -The research provides a method based on CPG to control a hexapod robot with multiple motion patterns, which can effectively overcome the difficulty of motion control simply by changing certain mathematical parameters of a nonlinear equation, such as frequency, phase difference and offset angle, so as to realize the gait transitions. Also, using such a robot to probe the distribution of snails offers another way to tackle this laborious job, especially in some odious terrains, which will hence broaden the application of AmphiHex-I to vector surveillance in the fields of public health.