The authors consider the problems of interaction of operators and robots, learning and telecontrol of the groups of different-function (heterogenic) robots. The learning-by-showing method is used ensuring recognition of the natural movement's shape of a human operator. The preliminary learning of the gesture and "signal" pose recognition of both human and android robot is based on (teaching) learning-by-showing method. The Model of Movements Shape (MMS) is based on М. Minsky theory of the frame knowledge representation proposed for memorizing of the semantic-topology frames system of the typical shapes of the working movements such as the typical scenario of a robot's behavior. Some typical objects {B} of MMS are: simple robot movements, fragments of a complex shape of the robot movement trajectories or sequences (cyclogrammes) of the working robot operations (fragments of scenarios of a robot behavior). Semantic features of {B} are determined as frame with name, parameters' fields, attributes of relations, application methods, etc. For organizing of the frame hierarchy structure, the connecting cross links within the frame structure are implemented during the learning, such as: "Parent attributes" A p, "Inheritance attributes" Ac, "Neighbor attributes" AN, "Method attributes" Am and other types of links. The number of the elementary objects {B} (working movements or sequences of the robots' operations) memorized at MMS will be significantly increased with enlarging of the volume of the robot tasks, but the mass of this objects will be considerably less than the area of the robot behavior scenarios (missions) in a real dynamical environment. The experimental research done with a virtual game modeling and real android football robots of NAO teams proved that the proposed learning-by-showing method and the telecontrol methods based on MMS Knowledge Base can be effective for the autonomous robot team telecontrol and correction of the robots' behavior in a real time mode. The based on IPv4 protocol programmable telecommunication of human-operator with a robot team can implement a real-time monitoring of the informative status of the robot telecommunication net for an adaptive control of the human-robot message parameters and/or a route track for the data transmission chain to each robot. The proposed methods of learning-by-showing and telecontrol using MMS Knowledge Base are expected to be used for assistant robots and anthropomorphic service robots, including for the medical or telemedicine purposes.