In areas without mapped streets, difficult landscape and challenging vegetation common path planning algorithms as such used for example in navigation systems do not work always probably. For these cases, when due to a good sensor data fusion a larger local map is available Mixed Integer Programming can be used as a method for trajectory generation for unmanned autonomous or semiautonomous systems. As several tasks for unmanned systems in harsh and difficult environments like border patrol, search and rescue missions, reconnaissance and surveillance, camp security and transport convoy missions require a group or formation of systems, as UAVs, UGVs or a mixture of them, the description of various formations and optimization of trajectories with respect to stability of formation and changing between different formations will be shown also. Additionally for the local collision avoidance an algorithm with a smaller planning horizon becomes necessary, to ensure the online capability; therefore here a Fuzzy Control approach will be shown. As these algorithms are going partially beyond the actually realizable capabilities of on the marked available systems, the state of the art in European robots for harsh and difficult environments will be shown to explain the missing links between theory, simulation and real world systems. For the explanation of what is already possible and how the future might go on, already realized systems using several parts of the here shown algorithms will be explained together with links to European projects in this field of robotics.