Autonomous OMAV Flight in Confined Spaces

Type of position: 
Master thesis
Short abstract: 
In this project the student will integrate the Voliro omnidirectional multi-aerial vehicle (OMAV) recently spun off from the lab with a navigation stack. The aim is autonomous flight in confined spaces such as doorways and narrow passages. Good results may see use in a high-profile research project.
Description: 

The Voliro [1] is omnidirectional and can tilt to fly in both horizontal and vertical configurations (see image). This allows the vehicle to squeeze through doors and narrow passages that are normally too narrow for standard aerial vehicles of this size. For this purpose we are equipping it with a state-of-the-art sensor and navigation stack developed by the lab for the DARPA Sub-T challenge [2-3], including high-res LIDAR and cameras.

While the sensors will already be mounted on the platform, they need to be integrated with the navigation stack. The navigation stack will also require modifications to work when the Voliro flies sideways in such a configuration. In particular, planning trajectories through confined spaces such as doorways, while taking into account the Volio’s unique flight configurations, in addition to sufficient LIDAR visibility for safe navigation. This would likely build on some existing UAV motion planner (see e.g. [4-6]). The entire design needs to be validated by real flight tests with the Voliro.

References:

[1] https://www.voliro.com

[2] https://www.subtchallenge.com

[3] https://youtu.be/sXVIUbe-RxA

[4] Helen Oleynikova, Zachary Taylor, Alexander Millane, Roland Siegwart, and Juan Nieto, “An Open‐Source System for Vision‐Based Micro‐Aerial Vehicle Mapping, Planning, and Flight in Cluttered Environments”. Journal of Field Robotics, 2020.

[5] Boyu Zhou, Fei Gao, Luqi Wang, Chuhao Liu and Shaojie Shen, “Robust and Efficient Quadrotor Trajectory Generation for Fast Autonomous Flight”, IEEE Robotics and Automation Letters (RA-L), 2019.

[6] Dharmadhikari, Mihir, Tung Dang, Lukas Solanka, Johannes Loje, Huan Nguyen, Nikhil Khedekar, and Kostas Alexis. "Motion primitives-based path planning for fast and agile exploration using aerial robots." In 2020 IEEE International Conference on Robotics and Automation (ICRA), 2020.

Work Packages: 
  • Integrate the mapping software (ROS/C++) with the new Voliro sensor setup for both flight configurations
  • Validate mapping via autonomous flight tests
  • Literature review of motion planning in confined spaces
  • Evaluate some existing planner(s) (e.g. [4-6])
  • Implement motion planning strategy for flight with multiple flight configurations (visibility aware)
  • Validate by flying the real Voliro through confined spaces (e.g. doors) using no external sensors
Requirements: 
  • Highly motivated student
  • For mapping part: C++, any experience with ROS, mapping or LIDARs helpful
  • For motion planning part: C++, some experience related to motion planning, for example control/trajectory optimization, obstacle avoidance or planning algorithms. Prior experience with ROS helpf
Contact Details: 

Please send your CV and transcripts to: Olov Andersson (PhD) olov.andersson@mavt.ethz.ch and Marco Tognon (PhD) mtognon@ethz.ch

Supervisors : 
Dr. Marco Tognon
Dr. Olov Andersson