Rui Huang - PhD Thesis Proposal
PhD Thesis Proposal Examination: 3D Modeling of Urban Scene Using Quadrotors - Rui Huang
Unmanned quadrotor aircraft are rapidly becoming affordable, safe and maneuverable, which makes them not only accessible to professionals, but also hobbyists and novices. Especially when the drones are equipped with high-resolution cameras, they can be used in many interesting applications such as aerial cinematography, object tracking and 3D modeling. Currently, people can reconstruct a 3D model of an urban area using the image data captured by cameras mounted on a quadrotor. The drone and the mounted cameras have to be controlled manually by experienced pilots with much skill and dexterity. Moreover, it is difficult to determine if the captured dataset is sufficient or not to produce a complete 3D model with high quality. People have to manually examine the reconstructed model from the current dataset and repeatedly send quadrotors to capture more data. These make the 3D modeling using quadrotors a challenging and time consuming task which is inaccessible to novices.
In this proposal, we firstly introduce an approach to fast 3D modeling from images. An initial 3D model can be obtained from the image data captured using a predefined flight plan. By assessing the model using an evaluation scheme, the system can recognize incomplete or uncertain regions. A set of waypoints are then generated and used to guide the quadrotor to capture more data for improving and completing the model. Our fully autonomous system is designed to make 3D modeling using quadrotors quick and easy. We have demonstrated our system on synthetic data and the integration with real platforms are in progress.
Ph.D. Examining Committee:
Dr. Ping Tan, Senior Supervisor
Dr. Richard Vaughan, Supervisor