Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Tag based ground truth for evaluating SLAM across several rooms/outdoors
Develop a tag-based system to provide an absolute (ground truth) position estimate.
Keywords: april aruco tag SLAM computer vision robotics
We deal with developing algorithms that can recover a precise camera trajectory estimate from the images it records. To evaluate these algorithms, one typically needs ground truth that tells how the camera has ACTUALLY moved (having a ground truth also has other useful applications, such as machine learning, characterising the scene, ...). Ground truth is typically obtained from a motion capture system. These systems typically have the limitation that they are constrained to a single room. However, there are systems around that enable ground truth generation from observing visual tags with known poses (e.g. RCARS: https://arxiv.org/pdf/1507.02081.pdf ).
We deal with developing algorithms that can recover a precise camera trajectory estimate from the images it records. To evaluate these algorithms, one typically needs ground truth that tells how the camera has ACTUALLY moved (having a ground truth also has other useful applications, such as machine learning, characterising the scene, ...). Ground truth is typically obtained from a motion capture system. These systems typically have the limitation that they are constrained to a single room. However, there are systems around that enable ground truth generation from observing visual tags with known poses (e.g. RCARS: https://arxiv.org/pdf/1507.02081.pdf ).
In this project, you will deploy a tag-based ground truth system in our offices, based on an existing ground truth estimation system. Once the system is deployed, you will investigate how robust it is and how to make it more robust and/or more general.
In this project, you will deploy a tag-based ground truth system in our offices, based on an existing ground truth estimation system. Once the system is deployed, you will investigate how robust it is and how to make it more robust and/or more general.
Titus Cieslewski ( titus at ifi.uzh.ch ) Required skills: Linux, experience in ROS or a very strong ability to learn, C++/Python.
Titus Cieslewski ( titus at ifi.uzh.ch ) Required skills: Linux, experience in ROS or a very strong ability to learn, C++/Python.