Real-time 3D modeling using swarm UAVs

How can we utilize multiple cameras in parallel in order to do faster and better 3D modeling of a site? Moreover, is it possible to have swarm based algorithms that dynamically steer and direct the UAVs in response to desired 3D model properties?

In this assignment we propose to develop a simple simulator for the scene and the obtained 3D data points of the cameras, in order to focus on swarm behavior guiding the construction of the 3D model. These guiding algorithms could be classic swarm algorithms, game theoretic approaches or some new algorithm. Other interesting aspects that can be addressed in this assignment are rapid anomaly detection and cross sensor-domain comparison at 3D model level. This project could also be integrated with augmented reality or virtual reality concepts. Real-world validation tests can be conducted using FFI’s swarm UAVs.

 

FFI is researching swarm systems, both from a theoretical perspective and in terms of practical applications development (https://vimeo.com/170946568). Our goal is to operate and understand continuous swarm operation for area monitoring using multifunction RF, electrooptical and acoustic sensors onboard. In this cross disciplinary research we are interested in working with all kinds of talented students, ranging from sensor students to more theoretical students in artificial intelligence methods. In the following some themes for master projects are suggested. However, we are open to all aspects of swarm research. So, if you are interested in swarm you should talk to us and we will find a project that is in line with your research interests.

Publisert 1. feb. 2017 09:15 - Sist endret 1. feb. 2017 09:45

Veileder(e)

Omfang (studiepoeng)

60