10 predictions about AI and advanced robotics in CEA

Photo © Adobestock

While technical and economic factors will moderate the pace of change, AI and advanced robotics are expected to influence nearly every aspect of CEA operations. The Resource Innovation Institute assembled the AI & Advanced Robotics Working Group (AIRWG) to assess their current state and forecast their development. Here are 10 predictions about how the industry will be impacted.

1. Production efficiency and crop yields will increase, with some hopeful expectations for profits, too.

The Autonomous Greenhouse Challenge hosted by Wageningen University has shown that autonomous control can increase the yield of cucumber by 12%, decrease the use of energy in cherry tomato by 20% and increase the net profit in lettuce by 28%.

Another greenhouse study pitting AI against humans showed an increase in tomato yield by 10% and profit by 92%.

An industry report of a commercial trial states an increase in greenhouse mini cucumber yield of 20%, with a 15% reduction in energy use with AI.

A simulation of a strawberry harvesting robot indicated it was five times slower than human harvesters but overall more productive, since it could operate 24 hours per day.

According to subject matter experts, autonomous climate and irrigation management is evolving rapidly and is now used in more than 100 greenhouses worldwide. Several industry reports indicate greenhouses can expand their size using the same number of growers/managers. Labor management and synchronizing crop production with market demands will also improve.

An obvious postscript to the metrics above is that your results may vary. The AIRWG consistently noted that technology and the adoption rate should always fit the operator’s business model.

2. Breeding will make automation possible, just as it has done in other areas of horticulture.

Due to their crunch, romaine lettuce cultivars became a staple for automation during the demand explosion for pre-cut bagged salads in the late 1990s and early 2000s. The rise of tomatoes-on-the-vine and snack tomatoes drove production away from the field and into automated high-wire greenhouses. Everbearing (day-neutral) strawberry cultivars aligned with indoor year-round production cycles.

There are examples from outdoor horticulture. The transition from hand-picked blueberries to machine-harvestable varieties was driven by firmer-skinned cultivars that could be shake-harvested without fruit damage.

At the same time, optical sorting machines graded the berries automatically based on color, firmness and ripeness. Introducing the ‘Honeycrisp’ apple variety transformed orchard management, requiring higher precision, better labor planning and more scientific growing methods.

While challenging to grow, its high consumer demand and premium pricing justified these extra efforts, ultimately pushing the apple industry toward more advanced and controlled orchard practices.

3. CEA operations will require more technology infrastructure.

The most automated facilities will need servers, high-speed networks, additional sensors and processing hubs for nearby sensors. Robots that collect visual data will need Wi-Fi receivers in central aisles to offload the data collected by robots.

Greenhouses with critical IoT systems will need reliable cloud connections and backup systems. At least until robots become more flexible, navigation will require induction lines embedded in the aisle floors, as well as installation of lidar (light detection and ranging) systems or RFID beacons.

4. Resource efficiency will improve.

Citing experience in European CEA operations, members of the AIRWG estimate that AI-enabled climate control systems could reduce energy consumption by 30% to 40%. One member confirmed that the savings exceeded the extra energy the AI requires, such as that from the data servers.

Supporting this are three recent academic studies modeling the performance of indoor plant factories with and without AI-enhanced controls that yielded energy savings of 9%, 25% and 32% with AI. A 2023 greenhouse study using data from the Autonomous Greenhouse Challenge indicated AI reduced energy consumption, though its effect on CO2 emissions and water usage was less impactful.

Contrary to these findings, a report modeling indoor plant factories indicated that energy savings using AI methods were one-tenth the savings of installing other equipment upgrades. Even so, the ROI of the AI outperformed the traditional methods.

The AIRWG members also reported that AI-enhanced greenhouses could be designed without the “fear extra” capacity (10% to 30%) built into the design specs of heating and cooling systems. This would save energy operating the systems.

Integration of lighting control with utility providers for day-ahead market pricing information, as modeled by researchers from Cornell and Rutgers, indicated cost savings of up to a remarkable 80% compared to traditional lighting control.

Beyond energy management, integrated sensing and control systems enable more precise irrigation and nutrient delivery based on actual plant needs rather than predetermined schedules. Early detection of pests and diseases allows for more targeted application of crop protection products, reducing pesticide usage. Robotic sprayers have demonstrated lower pesticide use rates.

Data-driven cultivation can reduce waste. The ability to track and analyze detailed production metrics supports continuous improvement in harvest timing and resource inputs. Cultivation is better aligned with market demand, reducing food loss and storage while improving inventory management.

5. Data management and security will become significant responsibilities.

Integrating data across different platforms and systems requires attention to compatibility and standards. And there is no playbook yet. Sensor calibration and machine-learning anomaly detection will be paramount for yield prediction tied to meeting contract obligations.

Data security breaches can result in ransomware attacks and attempts to steal intellectual property or compromise the nation’s food supply. Hackers are not just remote but may also physically attempt to enter the facility.

Along with the security of the IT systems, all employees will need to be trained to avoid being manipulated into revealing sensitive information in person, by phone or by email, referred to as social engineering attacks.

6. Product quality will improve while losses decrease.

Certain disease infections and abiotic disorders, such as nutritional deficiencies, will be detected by computer vision before visibility to the human eye, allowing faster corrective action and reduced product loss.

This capability improves consistency in sorting and provides valuable feedback about production conditions that affect quality. Automated handling can reduce product damage and contamination risks of human handling.

More consistent growing conditions and better harvest timing improve product quality, while comprehensive data collection enhances traceability throughout production.

7. AI will better link cultivation with the supply chain and the customer.

Real-time production data sharing will support better coordination with customers and distribution partners. Automated grading and sorting systems will help ensure consistent product specifications, while improved production forecasting will enable more efficient logistics.

Digital integration extends across multiple growing locations and with supply chain partners. This enhanced connectivity enables more responsive supply chains while supporting better traceability and quality assurance.

8. Robot-ready cultivation systems will slowly begin to emerge.

This change represents both the most significant risk and reward. After all, one takeaway of the last five years is that the most successful CEA operators were farmers first who did not let technology get too far ahead.

One or more start-up operations could break through with a robot-friendly cultivation system with competitive unit economics. But more likely, the race will not be the swiftest. Existing high-tech operations with the luxury of R&D cultivation space will (or perhaps have already) partner with robotics providers and university researchers.

Their growers and accountants will drive the transition to new cultivation systems, ones that can be effectively scaled. Once fully implemented, these operations should be able to expand their market share.

Two examples of new cultivation systems for tomatoes are in the works, both using AI, robotics and plant transport automation. Both follow the principle of moving plants rather than moving humans.

A cherry tomato cultivation system reimagines production with compact plants and shorter growing cycles designed for robotic maintenance and harvesting.

In the other case, hydroponic vines are lifted from their hydroponic solutions and transported through imaging equipment, leaf strippers and harvesters. In both cases, faster life cycles and crop turnover reduce the risk of pests and diseases. Reduction in human handling reduces the risk of foodborne disease contamination.

9. AI-enhanced climate control will act like guardrails.

Rather than replacing growers’ knowledge, automation technologies are becoming tools that enhance their growing capabilities. AI-driven climate control will predict outcomes, suggest strategies and detect anomalies to prevent new growers from making critical mistakes, much like advanced driver assistance systems protect student drivers.

AI climate control also allows junior growers to operate a high-tech greenhouse with limited assistance. Companies will scale faster, provide attractive jobs and reduce the strong dependency on the implicit knowledge of a few senior growers.

10. Facility site selection criteria may change.

Automation technologies may influence facility siting decisions by reducing dependence on local labor availability and enabling operation near renewable energy. The ability to monitor and control multiple facilities remotely could support new approaches to facility scale and distribution, potentially enabling more distributed production closer to markets.

Clustered CEA operations could generate a business ecosystem. The U.S. is behind the world in developing CEA clusters, where hundreds of acres devoted to protected cultivation co-locate in “farm parks.” These would attract an ecosystem of supply chain vendors and equipment providers, including AI and robotics firms.