Human-like intelligence requires human-like interactions with the world . ” –

The emergence of robot applications and its growing availability to nontechnical users implies the development of new ways of interaction between this kind of electronic devices and users. Human Robot Interaction (HRI) is a research area about the study of the dynamics involved in the interaction between humans and robots. It involves several knowledge fields such as natural language processing, computer vision, machine learning, electronics and even social sciences like psychology and human communication. HRI aims at the creation of natural interfaces between human and robots which are intuitive and easy to use without previous knowledge or training. The main goal of this Master Thesis is the development of a gestural interface to interact with robots in a similar way as humans do, allowing the user to communicate information beyond linguistic description of the task (non-verbal communication). In order to fulfill this objective, the gesture recognition application has been implemented using the Microsoft’s Kinect v2 sensor. Hence, a real-time algorithm is described to deal with two kinds of gestures which are described; the static gestures and the dynamic ones, being the latter recognized using a weighted Dynamic Time Warping method. Skeletal features are used to define both kinds of gestural sequences, having each gesture its own set of specific features. The Kinect based gesture recognition application has been implemented in a multi-robot case. So, a NAO humanoid robot is in charge to interact with the users and respond to the visual signals they produce. Moreover, a wheeled Wifibot robot carries both the sensor and the NAO robot, easing navigation when necessary. The system is currently able to recognize two gestures, one of each kind (static and dynamic). The dynamic gesture consists in a wave movement which the user salutes the robot; meanwhile the static one is a pointing to an object gesture. When performed, the robot looks for objects near the location which has been pointed, and tries to detect which is the object that the user was referring to, asking him or her about it, if needed. When the object requested by the user is recognized, the robot goes down the wheeled platform, approaches to it and shows it to the user. A broad set of user tests have been carried out demonstrating that the system is, indeed, a natural approach to human robot interaction, with a fast response and easy to use, showing high gesture recognition rates. Possible applications of this kind of systems to household environments are also discussed.