The study is part of the project “Real-time control of robots for the meat industry” which is undertaken by a consortium of universities and major players within the Danish meat industry including: University of Copenhagen, Technical University of Denmark, Danish Meat Research Institute, IHFood, Danish Crown, Danpo, Linco, Rose Poultry, TICan, and Robotcenter Danmark. A major part of the sub-project is focused on identification and 3D-localization of the legs of anesthetized pigs. Methods for accurate and robust partial 3D-shape modeling will be central. Data from standard cameras as well as infrared 3D cameras and use of techniques from Machine learning, such as the shape priors applied for human pose estimation in the Kinect playstations, will be considered. Other parts of the project will focus on how to hang up the pigs for slaughtering. Part of the sub-project may be to develop accurate and fast methods for visual servoing of the robot during the last phase of picking task. In general, the wishes for robustness and speed and requirements on animal ethics are considerable.
The applicant should have: – An MSc in Computer Science, Engineering, Physics or Mathematics – An MSc thesis within Computer Vision, Image Processing, and/or Machine Learning – Excellent programming skills, experience in MATLAB and/or C++ – A solid background in mathematics and/or statistics – Knowledge on depth image data analysis – A wish to combine theoretical method development with practical experiments – Courage to work together with industrial oriented people, and to spend part of the project time with the project partners.