We propose a system for detection and avoidance of obstacles based on the use of fuzzy logic. Basically, data coming from a ZED stereo camera are used for the decision making based on fuzzy predicates. Validation and testing of the algorithms are provided under the ROS (Robot Operating System), which runs together with Gazebo 7. The developed tool is an important step towards solving problems related to the techniques and methods or algorithms that can be used in mobile robots for detection and avoidance of obstacles and other objects on their way. The proposed system proved to be very robust in the tests carried out in the simulated environment.