Development of cooperative behaviour for multi-robots system (MRS) is an important aspect of robotic task achievement. This is because it is safer to use more than one robot to achieve a task for continuity measurement of task completion and it is also quicker to complete a task on time with two or more robots as against one robot. Continuity measurement here means if one robot breaks down, the remaining robots can continue and complete the task. Inspection of environments such as underground tunnels, underwater and industrial pipes are harsh areas that need exploring and exploiting in field robotics. The focus of this research is on using MRS to achieve pre-entry safety inspection of roof cracks and level of toxic gases in the underground terrain with the paramount aim of saving the lives of the miners from disasters. This paper presents the model for performance evaluations in terms of scalability. The use of cooperative behaviour to address the issues of multi-robot learning is also presented. We discuss the cooperative MRS behaviour framework and then describe the results from simulations. The evaluation/scalability of the model using two robots, three robots and four robots are investigated. The results reveal that our model can guide two, three and up to four robots to achieve efficient cooperative inspection behaviour in underground terrains with minimal cost of time and memory. The model will also work very well with any number of robots so long as the environment is expanded.