The bentonite-water mixture was selected as the substitute of seabed sediments according to the in-situ measurement data of sediments 15-20 cm deep in China’s ocean poly-metallic mining contract area and the soft seabed sediments could be simulated with certain proportion of the bentonite and water; besides, based on the theory on the interaction between the vehicle and ground and referenced to Bekker’s apparatus and related experimental methods, a scenario on the experimental system of the pressure-sinkage characteristics of interaction between the track of tracked vehicle and soft seabed sediments was designed. The pressure-sinkage experiments were performed with different dimensions of penetration plates. The “pressure-sinkage” model based on Bekker’s formula and correlation parameters were obtained to describe the corresponding characteristics of the seabed sediments and a smart calibration model on the pressure-sinkage characteristic of the track was established based on the function chain neural network, which could provide boundary loading conditions for simulation analysis of the tracked vehicle moving on the seabed. The bentonite-water mixture was selected as the substitute of seabed sediments according to the in-situ measurement data of sediments 15-20 cm deep in China’s ocean poly-metallic mining contract area and the soft seabed sediments could be simulated with certain proportion of the bentonite and water; besides, based on the theory on the interaction between the vehicle and ground and referenced to Bekker’s apparatus and related experimental methods, a scenario on the experimental system of the pressure-sinkage characteristics of interaction between the track of tracked vehicle and soft The pressure-sinkage experiments were performed with different dimensions of penetration plates. The “pressure-sinkage ” model based on Bekker’s formula and correlation parameters were obtained to describe the corresponding characteristics of the seabed sediments and a smart calibration model on the pressure-sinkage characteristic of the track was established based on the fun ction chain neural network, which could provide boundary loading conditions for simulation analysis of the tracked vehicle moving on the seabed.