Most of the existing screw drive in-pipe robots cannot actively adjust the maximum traction capacity, which limits the adaptability to the wide range of variable environment resistance, especially in curved pipes. In order to solve this problem, a screw drive in-pipe robot based on adaptive linkage mechanism is proposed. The differential property of the adaptive linkage mechanism allows the robot to move without motion interference in the straight and varied curved pipes by adjusting inclining angles of rollers self-adaptively. The maximum traction capacity of the robot can be changed by actively adjusting the inclining angles of rollers. In order to improve the adaptability to the variable resistance, a torque control method based on the fuzzy controller is proposed. For the variable environment resistance, the proposed control method can not only ensure enough traction force, but also limit the output torque in a feasible region. In the simulations, the robot with the proposed control method is compared to the robot with fixed inclining angles of rollers. The results show that the combination of the torque control method and the proposed robot achieves the better adaptability to the variable resistance in the straight and curved pipes. Most of the existing screw drive in-pipe robots can not actively adjust the maximum traction capacity, which limits the adaptability to the wide range of variable environment resistance, especially in curved pipes. In order to solve this problem, a screw drive in-pipe robot based on adaptive linkage mechanism the robot to move without motion interference in the straight and varied curved pipes by adjusting inclining angles of rollers self-adaptively. The maximum traction capacity of the robot can be changed by active adjusting the inclination of the variable resistance, a torque control method based on the fuzzy controller is proposed. For the variable environment resistance, the proposed control method can not only ensure just enough traction force , but also limit the output torque in a feasible region. In the simulations, the robot with the proposed contro l results is compared to the robot with fixed inclining angles of rollers. The results show that the combination of the torque control method and the proposed robot achieves the better adaptability to the variable resistance in the straight and curved pipes.