Among the many complex tasks in artificial intelligence, there is one, without the solution of which there can be no talk of any independent robot. It's a matter of planning. An important stage in any behavior is preparation for it, i.e. creating a plan for your action. However, in order to plan well, you need to have experience in performing these very actions - you need to learn what certain operations and sequences of actions lead to.
To increase the degree of autonomy of robotic platforms, it is necessary to create new methods for planning their interaction with objects and movement in space.
In this area, we are engaged in the integration of computer vision methods and classical symbolic planning methods, we work with simulation environments (Apollo) and real robots.