Bioinspired vision systems for automated harvesting

Summary: Automation of labour intensive horticultural processes such as fruit and flower harvesting is considered critical to securing the long-term future of the agri- and horticulture industries in the UK, yet technological challenges remain. This work addresses the crucial, and relatively poorly resolved, problem of using cameras and vision systems to perceive fruits and flowers in challenging outdoor environments. It will provide novel bioinspired approaches which will underpin the vision systems that drive robotic harvesting systems.  We will take inspiration from insects (e.g. bees, fruit-flies, aphids) that use visual cues invisible to humans (e.g. ultra-violet and polarised light) and thus potentially overlooked by engineers. Working closely with industrial partners we will collect a novel image database of key fruit and flower produce at various stages of their growth (and health) across weather, lighting, and housing conditions using a specially developed camera mimicking the vision system of insects.  Subsequent analysis, including bio-inspired image processing methods, will allow the fundamental methods employed by these expert foragers to be revealed. Understanding these principles will have immediate knowledge and commercial impacts aligned with AHDB strategic priorities. Specifically smart horticulture, through the design of new perception systems to underpin automated harvesting. The successful delivery of the project aims are significantly enhanced through the unique combination of research expertise from University of Lincoln’s Institute for Agri-Food Technology & Centre for Autonomous Systems with our horticultural industrial partners ensuring tight coupling of academic output with commercial application.