Robotic system for horticultural crop monitoring tested in Western Cape vineyards
The CSIR has developed a cost-effective platform to inspect and monitor horticultural crops on local farms. This automated, intelligent robotic system has significant potential to improve the monitoring, production, harvesting and processing of produce.
Following the successful development of a prototype that was tested in laboratory conditions, field trials of this precision-farming tool were undertaken to monitor vineyards in collaboration with the Department of Viticulture and Oenology and the Institute for Wine Biotechnology at the Stellenbosch University.
The sensors of the robotic system have been configured to estimate grape yield and to monitor plant growth and canopy health. The data provided will help farmers in the early identification of anomalies and potential hazards, allowing them to mitigate factors that could potentially cost millions in yield loss.
One of the key considerations in developing the technology was durability, as the harsh South African climate and landscape have in the past prevented advanced robotic technologies of this nature from being utilised. The researchers developed a platform that was capable and durable by ensuring that it had more than the required weight carrying capacity and energy to drive the machinery. Computer-aided design allowed for the simulation of performance before construction and deployment of the platform.
The robot is able to navigate autonomously through the vineyard, using CSIR-developed data-sensor fusion techniques to combine the data from the different sensors. Using the CSIR-designed data-fusion algorithm, it is also able to perform path planning, obstacle avoidance and lane following. The robot development is part of a three-year programme evaluating and demonstrating robotics, automation and sensors in viticulture.
Work undertaken in this project is part of a broader initiative focusing on automation solutions to improve the competitiveness of horticulture in South Africa.