Using AI to optimise sugarcane planting practices

Compared to whole-stick planters, growers using billet planters typically plant three to four times more seed cane to compensate for:
1. Uncertainties as to the viability of the seed cane which is reduced due to damage through the harvest-plant process and the planting of less-viable single node billets.
2. The compounding uncertainties in the measurements used when planting (bin weight, weight per billet, eyes per billet and percentage viable billets).
3. The mass-flow control mechanisms used offer little control with measured rates varying by more than 100% within a few metres.
4. Poor billet to soil contact can occur and is made worse by uneven billet distribution along the row. When clumping of planted billets occurs, adding more billets further exacerbates this problem.

Poor planting can lead to reduced yields as demonstrated by our Ballina trial where the mean gap between emerged plants was 59 cm which indicates a 30% loss in the yield potential for that field. While the trial field was not planted under ideal conditions, it demonstrates the potential impact. It is under non-ideal planting and early establishment conditions where poor planting is likely to significantly impact yield.

Typically, a grower must wait for the planted billets to establish before gaps become visible and then it is too late in the season to remedy. To compensate, a lot more billets are planted to try and avoid gaps as the operator cannot accurately:
1. Assess the quality of the seed cane.
2. Assess the number of eyes being planted per metre.
3. Control the feed of the planter for a uniform distribution down the furrow.
4. Control billet placement and associated bridging of billets.
Because growers/contractors cannot accurately measure actual planting rates, they continue to significantly overplant.

Acknowledgement
This project is fully funded by the Department of Primary Industries and managed by Sugar Research Australia.

This project expands upon prior collaborations on multiple projects between Griffith University and industry, in particular:
• With Davco as part of the ARC Industrial Transformation Research Hub focused on enhancing farming productivity and disease prevention.
• With the SRA and Sunshine Sugar, ‘Billet Quality Assessment’ Small Milling Research Project.

The project aims to improve planting by developing a System with AI capabilities that can measure billet planting rates and assess billet viability. This system can be used at either:
• The seed cane harvester to provide early notification of seed cane quality issues, or
• On the planter, with the capability to provide real-time feedback included the option of closed loop planter control.

The project will explore how this real-time information can be used to improve planting with the primary objectives of a more uniform strike with fewer gaps whilst using less seed cane. We will investigate:

1. Reducing machinery and agrochemical inputs and associated environmental benefits.
2. Monitoring of seed cane quality at the harvester.
3. Impact of seed delivery chute configurations.
4. Uniformity of delivery across rows for multi-row planters.
5. Impact of germination potential of rubber versus steel rollers.
6. The feasibility of closed-loop control of the billet feed.
7. Relative performance of different planting mechanisms.
8. Agronomic impacts of changes to planter configurations.
9. Providing manufacturers with immediate quantitative feedback to improve their designs.
10. How the system can provide industry with a tool for ongoing research.
11. How the system can be used for more effective demonstrations.