One of the most important components of almost any automated agricultural technology is a reliable perception system providing information about crop, plants and the entire farm environment. For example, a vision system designed for recognising strawberry fruit can be used for plant health monitoring, yield forecasting but also for guiding a robotic arm able to pick individual fruit. Development of effective and efficient vision systems for soft fruit industry is challenging, however, due to numerous challenges including variable lighting conditions, reflections, occlusions, the non-rigid structure of the strawberry plants and relatively small size of the fruit.
In this talk, I will present recent advances in the field of soft fruit perception undertaken by the University of Lincoln team, addressing problems of detection, localisation and tracking of soft fruit together with the assessment of their properties such as ripeness and quality. The provided examples will be illustrated by practical robotic applications for the soft fruit industry deployed in real farms.