AASS Seminar - Sensing challenges in Agri-robotics or how to make veggies smile to the camera

The research centre AASS arranges a seminar with Polina Kurtser, AASS, Örebro University.

Abstract

Current state of the art in agri-robotics has demonstrated proof of concept in field conditions with operational feasibility achieved. The goal set by researchers in the early 90's of solving the whole loop needed for successful robotic operation in the field - has been accomplished for many case studies for harvesting, de-leafing and pruning robots. Several of those robots have presented very advanced capabilities, bringing the day where agri-robots will be available on the market closer than ever before. The main obstacle in current R&D is taking the products from proof-of-concept levels to financially feasible products to achieve market penetration.

In this talk I will talk about some of those challenges addressed by me and my colleagues in the recent years, including learning and sensing issues, planning of active sensing and harvesting operations, robotic data acquisition protocols and more. I will also raise the current challenges I'd like to research in the upcoming years such as semi-autonomous data labeling, generalization of segmentation and detection algorithms, and robotic data acquisition protocols.

Bio

Polina Kurtser started her post-doc at Örebro University in April 2019. She received her Phd in Industrial Engineering in 2019. Her research interests include active sensing, computer vision, and statistical learning algorithms for agricultural robotics and medical imaging. In the past years she was involved in the H2020 project SWEEPER aiming to bring the first autonomous pepper harvesting robot to the market and in the nationally funded project PHENOMICS aiming to develop an autonomous phenotyping robot.