Learning Motion Primitives for Agriculture

Institut

Lehrstuhl für Agrarmechatronik (WZW)

Typ

Masterarbeit /

Inhalt

experimentell / theoretisch /  

Beschreibung

Motion primitives have been used in many robotics applications to speed up path planning. We would like to test their usage for agricultural vehicles, like tractor-trailer combinations. We want to employ frameworks like Dynamic Motion Primitives, using both simulation and the real vehicle to find and validate the best primitives for the given application. Both traditional path finding algorithms and discrete reinforcement learning policies could then use the final motion primitives.

Voraussetzungen

• A Bachelor’s level university degree
• Knowledge in Robotics and Path Planning
• Interest in basic machine learning applications
• Experience with MATLAB and Simulink
• Basic Knowledge of Motion Primitives (e.g. from Prof. Althoff's AI course)

Möglicher Beginn

sofort

Kontakt

Lukas Pindl
lukas.pindltum.de

Ausschreibung