Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Long-Term Aerial Localization in Agricultural Environment
Digital environments, or digital twins, allow for design, prototyping, and testing in the virtual world before moving to the real world, thus accelerating development and reducing costs. A digital twin of a farm supports crop operations such as scheduling a harvest or predicting a yield, while agritech companies can develop farm automation robots using a digital twin. The goal of this project is to develop 3D Reconstruction and localization strategies that are capable to identify temporal invariant areas and properties in crop environments during the production season. The main target is to be able to match the same plants over time.
Keywords: 3d Reconstruction, Place Recognition, Aerial Vehicles, Deep Learning, Computer Vision.
The construction of digital environments is increasingly important in industries like aerospace, automotive, and AEC. Digital environments, or digital twins, allow for design, prototyping, and testing in the virtual world before moving to the real world, thus accelerating development and reducing costs. Furthermore, a digital twin enables decision-making and -evaluation in the virtual world before committing to actions in the real world. Digital environments will play an equally important role in future agriculture. A digital twin of a farm supports crop operations such as scheduling a harvest or predicting a yield, while agritech companies can develop farm automation robots using a digital twin. An agricultural digital twin differs from an industrial digital twin in the need for periodic monitoring of growing crops and conditions on farmland, and in requiring new methods for capturing vegetation geometry and modeling plant growth that exceed the state-of-the-art. As such, the goal of this project is to develop 3D Reconstruction and localization strategies that are capable to identify temporal invariant areas and properties in crop environments during the production season. The main target is to be able to match the same plants over time. For this, the dynamic aspects of the model will have to be taken in consideration. Geometric and Machine learning approaches will be necessary in this project. Simulation and real-drones will be used in the project. The student will have the opportunity to learn all aspect, since the capture of the data to the final temporal reconstruction. Both perennial and non-perennial crops will be considered in the project. In addition, this work will be carried out in cooperation with a international company. The student taking this project needs to be highly motivated, preferably with strong analytical skills, while experience in coding in C/C++ and Python, Computer Vision, and Deep Learning would be very beneficial. Main reference: Surber et al., Robust Visual-Inertial Localization with Weak GPS Priors for Repetitive UAV Flights, ICRA 2017
The construction of digital environments is increasingly important in industries like aerospace, automotive, and AEC. Digital environments, or digital twins, allow for design, prototyping, and testing in the virtual world before moving to the real world, thus accelerating development and reducing costs. Furthermore, a digital twin enables decision-making and -evaluation in the virtual world before committing to actions in the real world. Digital environments will play an equally important role in future agriculture. A digital twin of a farm supports crop operations such as scheduling a harvest or predicting a yield, while agritech companies can develop farm automation robots using a digital twin. An agricultural digital twin differs from an industrial digital twin in the need for periodic monitoring of growing crops and conditions on farmland, and in requiring new methods for capturing vegetation geometry and modeling plant growth that exceed the state-of-the-art. As such, the goal of this project is to develop 3D Reconstruction and localization strategies that are capable to identify temporal invariant areas and properties in crop environments during the production season. The main target is to be able to match the same plants over time. For this, the dynamic aspects of the model will have to be taken in consideration. Geometric and Machine learning approaches will be necessary in this project. Simulation and real-drones will be used in the project. The student will have the opportunity to learn all aspect, since the capture of the data to the final temporal reconstruction. Both perennial and non-perennial crops will be considered in the project. In addition, this work will be carried out in cooperation with a international company. The student taking this project needs to be highly motivated, preferably with strong analytical skills, while experience in coding in C/C++ and Python, Computer Vision, and Deep Learning would be very beneficial. Main reference: Surber et al., Robust Visual-Inertial Localization with Weak GPS Priors for Repetitive UAV Flights, ICRA 2017
- WP1: Literature review of existing state-of-the-art 3D reconstruction for agricultural environments,
- WP2: Build a simulation-based and real-life dataset using Lab-tools and drones.
- WP3: Development of an reconstruction pipeline capable of selection more stable features for long-term data.
- WP4: Evaluation of the proposed method and comparison against state-of-the-art approaches and report writing.
- WP1: Literature review of existing state-of-the-art 3D reconstruction for agricultural environments, - WP2: Build a simulation-based and real-life dataset using Lab-tools and drones. - WP3: Development of an reconstruction pipeline capable of selection more stable features for long-term data. - WP4: Evaluation of the proposed method and comparison against state-of-the-art approaches and report writing.
The student taking this project needs to be highly motivated, preferably with strong analytical skills, while experience in coding in C/C++ and Python, Computer Vision, and Deep Learning would be very beneficial
The student taking this project needs to be highly motivated, preferably with strong analytical skills, while experience in coding in C/C++ and Python, Computer Vision, and Deep Learning would be very beneficial
Please send CV and Transcripts to Lucas Teixeira, lteixeira@mavt.ethz.ch
Please send CV and Transcripts to Lucas Teixeira, lteixeira@mavt.ethz.ch
