Artificial intelligence shows promise in hydroponics farming

The precision farming technology sector is hard at work developing artificial intelligence-based solutions for growers, but such systems cannot simply be plugged into existing systems, says Jeanne van der Merwe.

- Jeanne van der Merwe

The principle of implementing artificial intelligence (AI) in hydroponic farming systems is simple: integrate the production and monitoring technology in a hydroponic system, link them to an AI algorithm for your chosen crop, train it to perform the necessary actions according to your required parameters, and enjoy the resulting improvements in yield and income.

Designing a custom AI solution is, however, a far cry from typing a query into a chatbot. While AI-run production systems have been built in research settings, they do have limitations.

“We often see AI requested where the underlying need is better sensing, data capture, monitoring, or rule-based automation,” says Shaun Lamminga, engineering executive at Farmtrace, an enterprise resource planning provider focusing on agriculture that has implemented a range of systems in South Africa focusing on data capture, verification, telemetry, and traceability.

On occasion, their systems have included selective AI use, particularly with regard to computer vision and forecasting.

Lamminga says often the problems clients want them to solve do not necessarily require AI applications. “In many cases, traditional control systems, thresholds, trend analysis, and well-designed workflows solve the problem more cost-effectively and with less operational risk. AI tends to add real value only once the fundamentals are already in place.

“In highly controlled environments, most operational gains come from accurate sensing, calibration, monitoring, and responsive control systems. These approaches can guide adjustments to production parameters very effectively without the cost and complexity of training and maintaining AI models.