Stockton University and the Atlantic County Institute of Technology (ACIT) have joined forces to track bee movements with drones and advanced imaging at Macrie Brothers Blueberry Farm in Hammonton, New Jersey, according to Stockton University’s announcement. The pairing puts a university and a technical high school on a question that is hard to settle by eye: are the bees a grower relies on actually working the farm’s blueberry bushes, or are they flying off to other fields?
As of October 27, 2026, this is a new collaboration with proposed applications, not a finished study. That distinction matters, and this article keeps it front and center. Drones are no longer exotic on American farms. Our earlier look at farm drones treating 16.4 million acres in 2025 showed how fast spraying and scouting flights have spread. Pointing that same hardware at insects is a different and harder problem, and it is where imaging and machine learning start to earn their keep.
Below, we lay out what the partners have confirmed, what the sources leave open, how the technology could work in practice, and what the project says about the direction of AI-assisted agriculture. We also compare monitoring methods, trace the history, and offer five predictions. Where a detail is missing from the announcement, we say so instead of filling the gap.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Stockton University and ACIT Announced
The announcement describes a collaboration between Stockton University and ACIT. The project will use drones and advanced imaging technology to follow bees across a Hammonton blueberry farm. The study site is Macrie Brothers Blueberry Farm, a working commercial operation rather than a research plot, which gives the data a real-world setting from day one.
Local radio also picked up the news. 97.3 ESPN carried a piece titled Atlantic County, NJ Drone Effort At Stockton University & ACIT, which shows the story reached a regional audience beyond campus.
The project may also yield practical data. According to the announcement, the work could inform where growers place beehives and show how pollination patterns differ across blueberry varieties. Those are modest, farm-level outputs. They are also exactly the kind of decisions a grower makes every spring, which is why this study looks more useful than a flashy demo.
The Bee Question at the Center of the Project
Blueberries depend on insects to move pollen between flowers, and growers often bring in managed hives to make sure enough pollinators are on site when bushes bloom. The hive is easy to place. What happens after the bees leave it is much harder to see. A forager can work the nearest rows, drift toward a neighbor’s field, or favor one variety over another because of bloom timing or flower shape.
The announcement frames the project around that gap. Researchers want to learn whether bees leaving their hives are pollinating the farm’s own bushes or traveling elsewhere. If the answer varies by variety or by hive position, a grower could move hives, adjust numbers, or rethink planting layouts. That is a cost question as much as a science question, because hives and the labor to manage them are line items on a farm budget.
Tracking an insect that small over open ground is a genuine technical challenge. A bee moves fast, spends much of its time near the canopy, and looks like a few pixels from altitude. Field staff can count bees on flowers by hand, but that method samples tiny patches and depends on the observer’s time and eyesight. Aerial imaging promises wider coverage and a record that can be reviewed later.
Why Drones and Advanced Imaging Fit This Problem
A drone gives a research team three things a ground observer cannot match. It covers a whole block in minutes. It repeats the same flight path on different days, so results compare cleanly. And it carries sensors, from ordinary color cameras to thermal units, that record data a person would never notice. For a question about where bees go, coverage and repeatability matter more than any single spec.
From turf thermal maps to bee tracking
The partners are not starting from zero. Andrew Nebl, an ACIT aviation teacher, said ACIT students previously flew infrared cameras on drones at Tuckahoe Turf Farm in Hammonton to build thermal maps showing which parts of the turf received more water than others. That job taught students to plan flights, capture consistent imagery, and turn pictures into a map a farm manager can read.
Bee tracking asks more of the same skills. The students already know how to fly a grid, hold altitude, and process a batch of images. What changes is the target. Instead of a slow gradient in soil moisture, the camera has to catch small, fast subjects against a busy background of leaves and flowers.
What the camera has to see
The sources do not specify the cameras, so any detail here is analysis rather than reporting. In general, imaging for small flying insects trades off three things: resolution, frame rate, and flight height. Fly high and you see more ground but lose the insect. Fly low and you resolve the bee but disturb it and cover less area. Teams usually tune that balance on test flights before committing to a protocol.
Where AI Models Enter the Picture
The headline framing of this story pairs drones with AI research. The Stockton announcement, as summarized in our research, describes drones and advanced imaging. It does not name an artificial-intelligence system, a model architecture, or a vendor. We treat the AI role as likely but unconfirmed. Any project that records many hours of aerial footage will need software to find bees in it, because no student team can review that volume of video by hand.
That software is typically a computer-vision model trained to detect small moving objects and, in a second step, to link detections into paths. Training such a model needs labeled examples, and labeling is the slow, expensive part. Our report on Nvidia Flora labeling robot video at $0.72 an hour shows how much the industry is trying to automate exactly that step, which could eventually lower the cost for small teams like this one.
Where the model runs also matters. A school or farm may prefer to process footage on a laptop or workstation instead of sending it to a cloud service, both for cost and for control over farm data. Readers curious about that setup can see our walkthrough on running a model locally in 13 steps. Detection models are far lighter than large language models, so local processing is realistic for this kind of job.
For now, the honest answer is that the project’s software stack is unknown. The partners have not published a model name, a training set, or an accuracy target. Until they do, claims that the farm is running a specific AI system deserve skepticism.
What Is Confirmed and What Is Not
Early coverage of any new partnership tends to blur the line between announced intentions and operating systems. The table below separates the two, using only what the Stockton announcement and its fact sheet support.
| Item | Status | What the sources say |
|---|---|---|
| Partners | Confirmed | Stockton University and the Atlantic County Institute of Technology (ACIT) |
| Study site | Confirmed | Macrie Brothers Blueberry Farm, Hammonton |
| Technology | Confirmed | Drones and advanced imaging to track bee movements |
| Scholarship honoree | Confirmed | Paul “Ginger” Macrie Jr., former owner of the farm |
| Farm owner voice | Confirmed | Matt Macrie Jr., a current owner, is quoted in the announcement |
| Prior ACIT drone work | Confirmed | Infrared thermal maps at Tuckahoe Turf Farm, per Andrew Nebl |
| Possible outputs | Confirmed as potential | Beehive placement and pollination patterns across blueberry varieties |
| Drone model and maker | Unconfirmed | Not specified |
| AI system or model | Unconfirmed | Not specified |
| Imaging specifications | Unconfirmed | Not specified |
| Budget and scholarship amount | Unconfirmed | Not specified in our verified material |
| Launch and completion dates | Unconfirmed | Not specified |
| Farm acreage | Unconfirmed | Not specified |
| Project already operational | Unconfirmed | The source describes a new collaboration and proposed applications |
The gaps in the lower half of that table are normal for a project at this stage. They also set the questions to watch as the work moves forward. When the partners name hardware, publish methods, or share early results, this picture will change.
The Macrie Family, the Scholarship, and What the Owner Says
The collaboration carries a personal thread. A scholarship tied to the project will honor Paul “Ginger” Macrie Jr., the former owner of Macrie Brothers Blueberry Farm. Matt Macrie Jr., one of the farm’s current owners, spoke about what the project means to the family. In Stockton’s announcement he said: “We’re proud to see our grandfather’s legacy continue to inspire research, innovation and the future of agriculture.”
He also explained the practical motive. Macrie said: “So, you can imagine the amount of bees that come into this area in a brief period. This study is meant to give us some data so we can be more economical.” That line is the clearest statement of purpose in the coverage. The goal is cost discipline for a working farm, backed by evidence instead of habit.
A third remark speaks to the wider farming sector. Macrie said: “For us, as farmers, as technology takes over this sector, we’re now looking for help, basically. And this is one door that’ll help us open to many more.” Read plainly, he describes a grower who sees technology arriving faster than a farm can absorb it and wants local students and faculty as partners. That framing carries weight, because adoption of farm tech often stalls on exactly this kind of support gap.
The announcement, as relayed in our research, does not include the scholarship’s dollar value or its eligibility rules in material we could verify, so we do not report either. The structure is clear enough: a named award linked to the project, with the family’s history attached to a forward-looking research effort.
ACIT’s Aviation Program Brings Flight Experience
ACIT’s contribution is not just labor. The school already runs drone work in its aviation program, and Nebl described a concrete example. In his words: “We’ve done a study with Tuckahoe Turf Farm in Hammonton using infrared cameras on our drones to create thermal maps and see what parts of the turf fields got more water than others.”
Two points follow from that statement. First, the school has handled agricultural clients before, so students know that a farm expects clear deliverables on a schedule. Second, the Tuckahoe project proves a repeatable pipeline from flight to finished map. A pipeline like that can be reused. Swap the thermal camera for a different sensor, change the flight pattern, and the same habits of calibration and documentation carry over.
For students, the appeal is straightforward. They get a real client, real data, and a path into a university through the scholarship tied to the project. For Stockton, the partnership connects campus researchers to a nearby technical school and to a farm that wants answers. All three sides gain something concrete, which is the best predictor that a small collaboration will survive past its first season.
Precision Agriculture in 2025 and 2026: The Calgary Context
The Hammonton project sits inside a larger shift. The University of Calgary described precision agriculture in 2025 as the integration of sensors, drones, satellites, and other data sources with AI models to produce site-specific recommendations for crops and livestock. Under that definition, a bee-tracking study is a natural fit. It collects aerial data, extracts patterns with software, and turns the output into a decision about where to put hives.
University of Calgary research also describes computer-vision models that process drone imagery to estimate crop yields, catch early signs of disease, and monitor large areas. Related uses include crop segmentation, soil-health assessment, and targeted fertilizer or pesticide application. Those are mature categories with years of published work behind them. Pollinator tracking is less developed, which is why a small, well-scoped pilot has value.
Calgary also highlighted Agrivanna, a student-built system that combines AI, satellite and drone imagery, and GPS-enabled livestock collars to automate grazing decisions. The parallel with Hammonton is worth drawing. In both cases, students sit close to the working end of the technology, and the product is a farm decision rather than a research paper. That student-to-farm pattern appears to be one of the more durable features of the sector in 2025 and 2026.
Drone Hardware Trends: What TechRadar’s Coverage Shows
Hardware is moving quickly too, though not always toward farm use. TechRadar covered a large drone display over Texas under the headline 10,000-drone fleet uses tray-based swarm tech to dominate the Texas sky. The story is about entertainment, not agriculture. It still shows how far coordinated fleet control and drone-in-a-box launch systems have advanced.
Those ideas could migrate to the field. A docked drone that launches on a schedule could, in principle, fly a pollination survey at the same time each day without a pilot on site. That would turn a one-off student project into continuous monitoring. Nothing in the Hammonton announcement says this is planned, so we flag it as a direction of travel rather than a feature.
Autonomy is spreading in other heavy-duty sectors as well. Our coverage of Bedrock Robotics raising $600M for excavator autonomy shows investors backing machines that work outdoors with little supervision. Agriculture is a close cousin of construction in that respect. Both involve uneven terrain, variable weather, and labor that is hard to find.
Competitive Comparison: Ways to Monitor Pollination
Growers and researchers have several ways to study pollinators. The table below compares the main options on qualitative terms. The ratings are an editorial assessment based on how each method generally works, not measured results from the Hammonton project.
| Method | Area covered | Detail per insect | Labor needed | Fit for tracking bee movement |
|---|---|---|---|---|
| Hand counts on flowers | Small plots | High, species can be noted | High | Weak, snapshots only |
| Hive-entrance observation | One hive at a time | Medium | Medium | Partial, shows departures not destinations |
| Fixed ground cameras | Narrow field of view | Medium to high | Low once installed | Partial, good for one spot |
| Drone color video | Whole blocks per flight | Low to medium | Medium | Strong if altitude and frame rate are tuned |
| Drone thermal imaging | Whole blocks per flight | Low | Medium | Uncertain for small insects |
| Satellite imagery | Regional | Very low | Low | Not suited to individual bees |
The pattern is easy to read. No single method wins on every column. Drones offer the best mix of coverage and repeatability for movement questions, but they give up per-insect detail unless flown low. That is why ground observation will probably stay in the toolkit as a cross-check, with drone data providing the wide view.
Historical Context: From Hand Counts to Aerial Sensing
For most of the history of managed pollination, growers relied on rules of thumb and on what they could see from the edge of a field. Hive placement followed experience, and results showed up months later as fruit set and yield. Research stations added structured counts, which improved the science but not the reach.
Drones changed the economics of observation over the past decade. Cameras got smaller, batteries lasted longer, and software for stitching and analyzing images matured. Schools started building aviation programs around the new tools, and farms began to hire flights for scouting, mapping, and spraying. ACIT’s turf thermal-mapping job is a small example of that wider shift.
Agriculture’s appetite for new biology and new data is growing alongside. Our report on Volare opening a 5,000-tonne insect plant and raising €7.1M shows capital flowing to insect-related farming ventures. Pollinator monitoring is a different business, but it draws on the same growing comfort with treating insects as a managed, measurable part of the food system.
Market Impact for Farm Tech and Education
Direct market impact from one farm study is small, and we will not inflate it. The project carries no disclosed budget in our verified material, and the partners have not announced a product. The signal is in the model. A technical school, a university, a donor-backed scholarship, and a working farm form a template that other regions can copy at low cost.
For vendors, the opportunity lies in the unglamorous middle of the stack: affordable cameras, labeling tools, and analysis software that a school can run without a data-science team. The question of who captures value from AI in agriculture is part of a bigger debate. Our write-up of McKinsey’s AI value curve and its $850B cost risk shows how sensitive projected gains are to cost assumptions. Small farms feel those costs sooner than large ones.
For education, the effect may be larger. Matt Macrie’s remark about a door opening to more help points at a workforce angle. Students who learn to fly, process imagery, and report findings to a farmer leave with skills that employers in agriculture and engineering already want. Readers who follow the wider field can find more in our AI and machine learning coverage.
Five Predictions for Drone and AI Pollination Monitoring
These are editorial forecasts based on the facts above. They are not claims from the partners, and none of them is confirmed.
- The partners will publish methods before results. A new school-and-university project usually shares its flight protocol and imaging choices first, then early observations after a bloom season.
- Hive placement will be the first practical output. The announcement names beehive placement as a likely benefit, so expect early findings to focus on where hives sit relative to blueberry varieties.
- A named detection model will surface. If the team uses AI to find bees in footage, a model or tool name will probably appear in any later report or presentation.
- Neighboring farms will ask to join. Macrie’s comment about farmers needing help with technology suggests demand from other growers once the first findings circulate.
- ACIT will use the project to build agriculture coursework. A hands-on farm client gives a technical school a strong reason to formalize agricultural drone training, though no such plan is confirmed.
If the first season produces a clean hive-placement recommendation, the project will have proved its worth even without a single headline-grabbing number. If it produces only footage and questions, the team will still have built a reusable pipeline.
FAQ: Drones, Bees, and the Hammonton Project
Who is running the bee-tracking project?
Stockton University and the Atlantic County Institute of Technology (ACIT) are collaborating on it. The study site is Macrie Brothers Blueberry Farm in Hammonton, New Jersey.
What technology does the project use?
The announcement says drones and advanced imaging technology will track bee movements. The drone model, imaging specifications, and any AI system have not been confirmed.
Is the study already underway?
That is not confirmed. The source describes a new collaboration and proposed applications, so readers should not assume operating results exist yet.
What could growers learn from it?
The data may help with beehive placement and show how pollination patterns differ across blueberry varieties. Both outputs are described as possibilities.
Who is Paul “Ginger” Macrie Jr.?
He is the former owner of Macrie Brothers Blueberry Farm, and the scholarship tied to the project will honor him. Matt Macrie Jr. is identified as one of the farm’s current owners.
Has ACIT done drone work on farms before?
Yes. Aviation teacher Andrew Nebl said ACIT students used infrared cameras on drones at Tuckahoe Turf Farm in Hammonton to make thermal maps of water distribution across the turf.
Does the scholarship have a confirmed dollar amount?
Not in the material we verified. We list the scholarship amount, project budget, and timeline as unconfirmed.
