Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Visual-inertial odometry and Event dataset for Drone Racing
Visual-inertial odometry and Event dataset for Drone Racing
Vision-based autonomous drone racing requires Visual-Inertial Odometry (VIO) algorithms that perform well on large optical flow and motion blur. Existing benchmark datasets often fail to address fast and aggressive trajectories as observed in drone racing or lack high-resolution ground-truth poses. This project aims at collecting an extension of the UZH-FPV Drone Racing dataset [https://fpv.ifi.uzh.ch/], previously used for international VIO competitions. The student will learn about VIO and event data collection in a large-scale position tracking arena, multi-camera-IMU calibration and time-synchronization of multimodal data, and will compare state-of-the-art VIO algorithms on this new dataset. Requirements: Experience with Linux, ROS, Python; CPP knowledge is a plus; Prior experience in quadrotor flight is a plus but not strictly necessary.
Vision-based autonomous drone racing requires Visual-Inertial Odometry (VIO) algorithms that perform well on large optical flow and motion blur. Existing benchmark datasets often fail to address fast and aggressive trajectories as observed in drone racing or lack high-resolution ground-truth poses. This project aims at collecting an extension of the UZH-FPV Drone Racing dataset [https://fpv.ifi.uzh.ch/], previously used for international VIO competitions. The student will learn about VIO and event data collection in a large-scale position tracking arena, multi-camera-IMU calibration and time-synchronization of multimodal data, and will compare state-of-the-art VIO algorithms on this new dataset. Requirements: Experience with Linux, ROS, Python; CPP knowledge is a plus; Prior experience in quadrotor flight is a plus but not strictly necessary.
The goal of this project is to collect a visual-inertial and event dataset for drone racing and curate the data to meet the standards of international VIO competitions.
The goal of this project is to collect a visual-inertial and event dataset for drone racing and curate the data to meet the standards of international VIO competitions.
Christian Pfeiffer (cpfeiffe@ifi.uzh.ch), Giovanni Cioffi (cioffi@ifi.uzh.ch)
Christian Pfeiffer (cpfeiffe@ifi.uzh.ch), Giovanni Cioffi (cioffi@ifi.uzh.ch)