Thesis supervisor: Kristian Arian Fenech
Location of studies (in Hungarian): Eötvös Loránd Tudományegyetem Abbreviation of location of studies: ELTE
Description of the research topic:
Effective human-machine collaboration needs effective communication and robust representations. The aim of this project is to investigate how machines can learn to perceive and understand complex and pragmatic signs from humans as they navigate and perform tasks in highly complex and dynamic environments. Machines in these cases will need to be able to estimate attributes which give rise to infomation regarding a persons internal state, their intentions and goals. As well as there external state, their motion now and in the near future. In addition, how human and machine can effectively communicate can be altered both by physical limitations enforced due to certain activities or environmental conditions, however the machine must not fail in its interpretation given this altered representations. This requires work towards the development of rich digitial representations of objects, environments and people which can be used and updated in real-time forming expressive digital-twins which may be used in a wide variety of use cases such as path planning and optimisation in factories, security and safety, and virtual training environments.
Required language skills: angol Further requirements: • Strong knowledge of linear algebra, machine learning and python programming.
• Knowledge of robotic systems is a bonus
• Knowledge in computer graphics and common tools such as Unity3D/Blender is beneficial but not essential
Login
