Parallel and distributed algorithms are largely investigated in HPC community. However new distributed computing schemes still need to be investigated to cope with the resource constrained embedded devices, wireless IoT communication characteristics and real-time requirements. For example, some specific works have been done concerning one of our topic of interest which is multi-robot exploration. We aim at studying original distributed computing schemes to efficiently manage the large amount of computations that is often required in CPS applications like exploration or monitoring. Here again, some works exist but we would like to extend our investigations on hybrid schemes, to combine centralized and distributed computing and take into account both computing resource and communication constraints. In the continuity of this topic, we study also self-adaptive algorithms for CPS robustness.