University of Canterbury: Autonomous 3D crop scanning for yield estimation, seeking trial partners
Opportunity for
- Vertically integrated producers in specialty tree crops (apples, cherries, stonefruit) and vineyards who want significantly improved yield estimate certainty
- Orchard and vineyard growers who actively monitor and predict yield and are open to trialling new precision technology
- Packhouses and fruit graders wanting more accurate pre-harvest data to better plan labour, packaging and sales
- Industry partners across the specialty crop value chain interested in validating and benchmarking new yield monitoring technology
Opportunity description
Industry challenge
Sampling a farm block periodically takes time and inherently costs more the more sampling you want to do. Despite this, sampling is hugely valuable for getting a better idea of the current state of your plants in terms of yield, vigour, hydration, disease and more.
Currently, standard practice for obtaining crop estimates involves sampling just 0.1 to 2% of the crop. Depending on the method of sampling, error rates are often between 15 and 25%. This error gets amplified significantly when extrapolating over a whole block. Even modern camera-based systems suffer from a similar range of inaccuracy, as they rely on a calibration stage completed by a farm worker.
Fruit graders and packhouses are consistently disappointed in the accuracy they are getting from the field and often receive yield estimates that cause a large over or under supply of labour, packaging, production line resources and fruit sales.
Current opportunity
Holocrop has developed an autonomous 3D scanning system that delivers precision crop data without relying on manual calibration or field workers. Holocrop is now seeking trial partners in Australia, specifically growers in specialty tree crops and vineyards who are looking for greater certainty in their yield estimates and are open to trialling a new approach that could significantly reduce the error margins they currently experience with traditional sampling or camera-based methods.
The trial will focus primarily on fruit count and size pre-harvest for yield estimation, using hyper-realistic 3D modelling and 3D AI detections to segment and extract detailed information from different plant structures on a per-plant basis. Holocrop will then conduct a study comparing any current yield monitoring and prediction systems deployed on trial farms against theirs to determine the return on investment for trial partners’ operations. The system is non-invasive and simply drives around your crop rows autonomously, capturing precision data.
Opportunity background
Holocrop is a spin-out from the University of Canterbury in New Zealand, founded by a team of research engineers who each bring prior industry experience. The technology has been developed as part of a multi-year, government-funded MBIE Endeavour research programme.
Holocrop's autonomous 3D scanning rovers are already deployed with research partners including Plant and Food Research in New Zealand, with further validation work underway with Washington State University. The team is currently participating in the Sprout Accelerator, New Zealand's largest agritech-focused venture programme.
The system has been validated across apples, cherries and grapes, and the team is actively building commercial partnerships to expand into Australia and the United States.
Potential other applications
Although the system will be calculating the count and size of fruit for this trial, it captures the whole 3D structure of the crop plants. This allows Holocrop in future to provide precise volumetric and sizing data for plant structures such as leaves, shoots, branches and trunks.
This capability opens up applications beyond yield estimation, including canopy management, pruning assessment, vigour monitoring and growth tracking over time.
The 3D data could also support automation and robotics applications, providing a detailed crop perception backbone that autonomous harvesting or pruning systems need to operate effectively.
For R&D teams, the system offers a way to quantify trial outcomes with a level of precision that is difficult to achieve with traditional field technician methods, at a fraction of the cost.
This makes it relevant not only for commercial growers but also for horticultural industry bodies, chemical companies, consultants, and contract researchers looking to improve the accuracy and efficiency of their field-based assessments.