Title: Object-level visual SLAM for plant modeling

Abstract:

A 3D model that can capture the finer details of the plant structure as well as the overall field statistics, plays an important role in automating agriculture. However, modeling and mapping an agricultural field is challenging due to dynamics, illumination conditions and limited texture inherent in an outdoor environment. We propose a pipeline that combines the recent advances in deep learning with traditional 3D processing techniques to achieve fast and accurate SLAM in vineyards. We merge image features with local shape information in 3D to classify plant structures and build a dense 3D map of the scene. In addition, we explore two particular applications of 3D modeling in vineyards: counting grapes, and mapping branches in the dormant season.

Committee:

George Kantor (advisor)

Katerina Fragkiadaki

Eric Westman