The design of a robot is tightly linked to it assigned task. A robot does not have the same adaptability as humans do. Therefore every robot is designed to achieve one specific task. An increasing alternative to this issue are modular robots. Since one can not achieve the task alone, it requires the help of others. In this thesis is studied the possibility of having one single ground robot managing the displacement task instead of implementing sensors and algorithms on each independent autonomous systems. The displacement robot should have the ability to plug itself to some others so that it can then drive them around at their desired places. This thesis work is especially focusing on this docking sequence between the displacement robot and the other module. The achievement of this docking sequence relies on the use of fiducial markers to extract the relative poses of the two systems. Furthermore, the noisiness of the obtained measure has been dealt with Extended Kalman Filters and Monte Carlo Localisation algorithms. The results obtained through the simulations shown that a MCL algorithm would provide better results for such systems. However the output relative poses were not precise enough for the system to actually plug connectors and ensure the communication between the two systems.