IPM 2.0: The role of AI and robotics in pest management programs
Since its inception in the 1950s, the concept of integrated pest management (IPM) has revolved around defining acceptable levels of pests and diseases and preventative cultural practices. Typically, this has involved extensive monitoring and record keeping, mechanical and biological control of pests & diseases and if necessary, responsible use of pesticides, preferably in limited areas. This core philosophy is still relevant today but since the 1950s, the size of operations has changed dramatically and with that, the size of the problem.
There was a time when we had fewer greenhouses and enough people to manage them, but the number of greenhouses has increased globally yet the level of human resources has not increased at the same rate. In these operations, knowledgeable humans cannot be everywhere at once and things can get out of hand. Couple that with limitations posed by plant viruses and the global pandemic, access to plants and farms is even more limited now than it was a few years ago.
This is where technology can help.
Over the past two decades, robots and automation, and more recently AI and machine learning, have entered farms and greenhouses. While some areas have fully embraced these technologies (ie, sorting machines based on computer vision, automatic trollies, climate computers, irrigation management) some areas are still in their early stages of research and development (eg robots for harvesting, de-leafing and plant lowering).
Technologies supporting IPM are more mature in comparison. As many know, monitoring is essential for IPM success, and many applications can now digitize human observations on the ground. The conventional practice consists of humans walking the crop, recording their location, and jotting down what they see on a piece of paper or simply memorizing and discussing them with their managers.
At a minimum, monitoring applications use mediums such as phones or tablets to record and digitize human observations. Some allow the user to snap and save a picture along with their notes. These apps can create a digital archive of human observations, and their records can be used to generate historical trends. They can facilitate some administrative aspects of IPM record-keeping, but these applications rely entirely on human input.
The second tier of monitoring technologies consists of sensors and cameras that passively or actively collect information from the crops to determine the health of the plants. These technologies range from high-resolution visible RGB to thermal, infrared, multispectral, hyperspectral and UV cameras, as well as climate, chemical, and electrophysiological sensors. The sensory data and imaging information is usually coupled with a machine learning/AI engine that either flags anomalies in the data sets or detects specific patterns or objects. Some of these platforms capture the data and send it to the cloud for further analytics and some use edge computing (data processed live on a chip) based on CPU or GPU to provide real-time analytics. All of these technologies require extensive training and a model monitoring platform to ensure their accuracy and performance are maintained. These cloud-based systems must address connectivity and bandwidth available on the farm as well as data security and privacy matters.
Most farms are in rural areas where high-speed internet is not available. Therefore, alternative solutions such as edge-based tech are a better fit compared to cloud-based solutions because they depend less on high-speed internet. Since, in edge-based technologies, the data stays on the farm, the possibility of hacking, data leaks and cyber-security breaches is significantly lower compared to cloud-based solutions where data is broadcasted beyond the farm.
The third tier of solutions combine the digitization of human observations with automatic data collection through a series of sensors and cameras. They offer a comprehensive and holistic description of verified pests and status of diseases in the greenhouse. Equipped with edge computing, these solutions can produce real-time risk alerts. Furthermore, these solutions can provide outbreak projections and predictions.
No matter what tier of technology is used for monitoring pests and diseases, one must use the incoming data to formulate a proper course of action and treatment to fix the problem. This is also an area where technology can significantly help growers. Using IPM analytics and additional measurements such as micro-climate, physiological state of the crop, pest and disease threshold levels, and age of the crop, AI systems can calculate the most optimized course of action that provides the best control for the lowest price.
But this is not the end. Even if accurate information is gleaned through good monitoring and the most optimized treatment is prescribed, one still needs to administer the treatment properly, in the right place, while following the correct instructions to achieve the desired results. This is yet another area where robots can help. Robotic sprayers have been in the market for more than a decade and now alternative solutions such as UV-C treatments, biopesticide sprayers and robotic systems for dispersal of biological control agents are becoming commercially available. We even have technologies that physically remove pests from the greenhouse, such as giant vacuums and moth-killing drones.
Combining all these technologies provides an opportunity for a new age of IPM programs where an end-to-end solution can be offered as a service. As a first step, growers should first determine their operational needs, their specific issues of interest, and their budgets. Service providers can then put together a package that consists of the right monitoring technologies, prescription algorithms and treatment platforms that take care of the problem. Human knowledge and influence will still be part of this new system, but the sheer size of the operation or its geographical location shouldn’t be limiting factors.
The human expert should still have the option to veto the machine’s recommendations and adjust them based on their insights or other considerations. The digital nature of these platforms and their ability to verify the effectiveness of a certain decision will allow growers to perform micro-experiments and fine-tune their knowledge while keeping a large crop clean and pest- and disease-free. The best way to ensure AI and robots can help and benefit your operations is to invest in a very knowledgeable IPM manager who can train and utilize these technologies and if necessary, overwrite their recommendations. In the same way that the most sophisticated and technologically advanced airplanes still need a human pilot to fly, the most advanced platforms for IPM still need a human supervisor.
Every year, more pesticide products are removed from horticulture’s IPM arsenal and every year, we are faced with new pests and diseases. Consumers are demanding pesticide-free food; Retailers have higher standards and use cosmetic damage as grounds for returning produce. And well, there is this pandemic that we all want to go away but won’t! In this environment, IPM tech is no longer a nice-to-have, but rather a necessity. Whether you go all in and invest in the full fleet of robots and AI or do the bare minimum of digitizing your practice through apps, it’s important to take action now to remain on top of your game and future-proof your operation.
Saber Miresmailli, PhD, is a former greenhouse grower, Canada Clean50 Leader, Top 40 under 40, award-winning biologist/IPM specialist and the founder and CEO of ecoation, a company with headquarters in British Columbia and offices in Ontario, Brazil, India , and Belgium that focuses on AI and robotics in horticulture. He can be reached at [email protected].