Aiming at the problem of obstacle avoidance during the operation of the tank surface, an improved fuzzy obstacle avoidance control algorithm is proposed. To ensure that the wall climbing robot can shorten the time to reach the target point, in addition to the distance between the wall-climbing robot and the obstacle, increasing angle between the wall-climbing robot and the target point as an input variable. Two variables of speed and angular velocity are used as output. After determining the domain and membership of each parameter, a fuzzy rule table is established. Defuzzification is the center of gravity of the application. Considering the complex variability of the tank environment, proposing a strategy of elegant downgrading obstacle avoidance control. The detection sensor can still cross the obstacle to reach the target point in the event of a temporary environmental failure. By comparing with the artificial potential field obstacle avoidance method in simulation, the improved fuzzy control method can avoid obstacles more effectively. And when the detection sensor is abnormal, the climbing wall robot can also be guaranteed to reach the target point. It demonstrates the superiority and reliability of the obstacle avoidance algorithm. Finally, carry out the obstacle avoidance experiment of the wall climbing robot and verified the feasibility of the algorithm.