Cost-effective weed control in cereals using vision guided inter-row hoeing and band spraying systems

N. Tillett

Abstract
Weed control is one of the most economically and agronomically significant problems for both conventional and organic cereal production. This project developed generic precision row guidance technology to benefit cereal producers through better targeting of both chemical and mechanical weed control inputs.

The technology is based on computer vision to detect and dynamically track crop rows. Previous work has been restricted to following only one drill bout, limiting work rate. In this project we overcame technical barriers to tracking multiple bouts and demonstrated this capability on a 12 m inter-row hoe spanning three 4 m drill bouts. Crop row spacing was increased to 250 mm to facilitate passage of 160 mm wide hoe blades. The experimental hoe comprised three sections each with its own camera and mechanism for independent lateral movement.

Field trials showed that under normal commercial conditions lateral error should not exceed 25 mm (S.D. 10 mm) at speeds of up to 10 kph. Performance was reliable under a wide range of crop growth stages and coped well with high levels of weed infestation.

Problems that did occur were usually due to inaccuracy of drill bout matching, particularly near headlands, where bouts converge or diverge beyond the lateral stroke (+/-250 mm). Strategies were successfully implemented for dealing with these eventualities, though for best performance extra attention should be paid to bout matching.

Two application scenarios were explored, organic and conventional production. For the latter, we implemented combined hoeing between rows and band spraying of selective herbicides on the row. This reduced herbicide input by 60%. Unfortunately, as a consequence of the very narrow (100 mm) spray band individual nozzle flow rates must be very low (10 litre/h). Conventional nozzles with appropriately small orifices are very prone to blockages.

Further investigation suggested that twin fluid nozzles would be more suitable for this application as satisfactory spray patterns can be generated at very low volumes (5 litre/h) potentially leading to overall application rates of 10-20 litre/ha. Although project objectives were restricted to the development and physical evaluation of engineering systems the hoe was used to treat approximately 5 ha of commercial organic spring wheat in 2004.

Results indicated that crop yield was 18% higher than a section of the same field drilled at 125 mm row spacing and harrowed. This result was not part of a replicated trial, and cannot be treated as significant in isolation, but it does highlight potential differences that are not well covered by contemporary research.

We speculate that the difference in crop yield, appearance and nitrogen levels (up from 12 to 13%) may be connected to mineralization of nitrogen rather than weed control. As part of the wider scope of this LINK project similar technology was used to guide four band spraying sections from a conventional boom sprayer operating in vegetables. Trial results indicated reliable performance albeit at a lower accuracy (22 mm S.D) due to the less stable platform.


Source
HGCA Project Report. 2005; (370): 25 pp. London, UK: Home Grown Cereals Authority.


Author Location and Affiliation
Silsoe Research Institute, Wrest Park, Silsoe, Bedfordshire, MK45 5HS, UK.


en français


Posted May 2007