The problem of navigating a formation of interconnected tethered drones,\nnamed STEM (System of TEthered Multicopters), in an unknown environment is\nconsidered. The tethers feed electrical power from a ground station to the\ndrones and also serve as communication links. The presence of more than one\ninterconnected drone provides enough degrees of freedom to navigate in a\ncluttered area. The leader drone in the formation must reach a given point of\ninterest, while the followers must move accordingly, avoiding interference with\nthe obstacles. The challenges are the uncertainty in the environment, with\nobstacles of unknown shape and position, the use of LiDAR (Light Detection And\nRanging) sensors, providing only partial information of the surroundings of\neach drone, and the presence of the tethers, which must not impact with the\nobstacles and pose additional constraints to how the drones can move. To cope\nwith these problems, a novel real-time planning algorithm based on numerical\noptimization is proposed: the reference position of each drone is chosen in a\ncentralized way via a convex program, where the LiDAR scans are used to\napproximate the free space and the drones are moved towards suitably defined\nintermediate goals in order to eventually reach the point of interest. The\napproach is successfully tested in numerical simulations with a realistic model\nof the system.\n