Per favore scegli almeno una figura professionale o una competenza




Figura professionale




Risultati di ricerca

Risultati per:

Thesis Work at Volvo Group - Leveraging multi-camera setup for semantic segmentation ( 2 Students )

Position DescriptionTo be prepared for a future where significantly increased flexibility is needed when it comes to e.g. number of produced product variants, Volvo Group are, together with some university partners, (e.g. Chalmers) developing a Generic Photogrammetry based Sensor System, GPSS. A camera-based sensor system with several cameras mounted in the factory ceiling. Applying artificial intelligence when interpretating the video stream the results can be used for e.g. controlling robots, tracking people and in general finding objects preventing the robot movementsSuitable backgroundWe are looking for two highly motivated students with great interest for computer vision and machine learning. We also believe that the students are eager to see their developed methods be put into practice at the Volvo pilot plant in Gothenburg. The following skills are of great value for this project:Deep LearningComputer VisionPython ProgrammingReading Scientific PapersDescription of thesis workThis work is a continuation of a previous successful master thesis that trained a semantic segmentation network for finding the free floor space. This was used for defining areas on the shop floor where Autonomous Transport Robots (ATR) could act and maneuver without the need of expensive on-board sensors. This thesis is about using the fact that there are more than one camera covering the same floor space for calculate the floor coverage of objects (i.e. as ground truth rather than the object’s pixels in the image). We believe it is likely that a multi-view setup would outperform a single camera for this task. Therefore, we should investigate if it is possible to predict a single floor coverage mask when using images from several cameras as input.Thesis Level: Master and/or BachelorLanguage: Final report in English. Swedish/English during the work.Starting date: January 2022Number of students: 2

VolvoAuto Sweden Sa • Stockholm, Sweden