Mobile devices such as smartphones, AR/VR glasses and robots will soon be required to fully understand the world around them in order to perform a large variety of applications, ranging from automatic inspection of infrastructure, to navigation in difficult environments. This understanding will come from a full and detailed digitalization of the world, which can be performed via 3D reconstruction algorithms. At Ericsson Research, our team has been developing state-of-the-art 3D reconstruction software which is revolutionizing the way devices see and understand the world.
We are currently looking for students to work on a Master Thesis project to investigate, design and test the boundaries of the modern 3D reconstruction algorithms. This is an opportunity for you to further explore the exciting field of 3D reconstruction, like Multi-View-Stereo (MVS) [1, 2], Structure-from-Motion (SfM) algorithms [3, 4], Visual-Inertial SLAM concepts [5, 6], and demonstrate your cutting-edge ideas, energy and enthusiasm to revolutionize future connected devices and machines.
The project will take place between January and June 2021.
The project will focus on one of the following areas:
1. Monocular 3D reconstruction with accurately estimated camera poses: The research question in this study is to find the optimal reconstruction algorithm, given that accurate camera poses are provided by an additional sensor.
2. 3D reconstruction via monocular and depth sensors: The work will focus on investigating the optimal fusion of 3D information from depth sensors and a monocular camera in order to obtain accurate 3D models with improved speed.
- Strong interest in computer vision, in particular 3D reconstruction
- Knowledge in computer vision, signal processing and optimization
- Ability to formulate technical problems and solve them independently and in groups
- Strong analytical skills and ability to acquire new knowledge, apply it, and spread it to others
- Experience in C/C++ and Python
- Strong communication skills in written and spoken English and good presentation skills
 Y. Furukawa and J. Ponce, “Accurate, Dense, and Robust Multi-View Stereopsis,” IEEE Trans. Pattern Analysis and Machine Intelligence, 2010.
 J. Schonberger and J. Frahm, “Structure-from-Motion Revisited,” in Proc. Conf. Computer Vision and Pattern Recognition, 2016.
 T. Schneider, M. Dymczyk, M. Fehr, K. Egger, S. Lynen, I. Gilitschenski and R. Siegwart, “maplab: An Open Framework for Research in Visual-inertial Mapping and Localization,” IEEE Robotics and Automation Letters, 2018.