Massachusetts Institute of Technology student Huan Liu of Shanghai, China, positions a robot gardener near a tomato plant while demonstrating its capabilities in the Artificial Intelligence Laboratory on March 18 on the school’s campus in Cambridge, Mass.

These gardeners would have green thumbs — if they had thumbs.

A class of undergraduates at the Massachusetts Institute of Technology has created a set of robots that can water, harvest and pollinate cherry tomato plants.

The small, $3,000 robots, which move through the garden on a base similar to a Roomba vacuum, are networked to the plants. When the plants indicate they need water, the robots can sprinkle them from a water pump. When the plants have a ripe tomato, the machines use their arms to pluck the fruit.

Even though robots have made few inroads into agriculture, these robots’ creators hope their technology eventually could be used by farmers to reduce the natural resources and the difficult labor needed to tend crops.

Last spring, Daniela Rus, a professor who runs the Distributed Robotics Lab at MIT, began a two-part course. In the first semester, the students learned the basics of creating and using robots. By the fall, the students were ready to have robots tackle a real-world problem. Rus and Nikolaus Correll, a postdoctoral assistant in Rus’ lab, challenged the students to create a “distributed robotic garden” by the end of the semester.

The 12 students broke into groups, each tasked with solving a different problem, such as creating the mechanical arm needed to harvest the tomatoes or perfecting the network that let the plants and robots share information.

By the end of the fall term, the “garden” inside Rus’ lab was green and growing.

Now there are four cherry tomato plants nestled into a plywood base covered in fake grass. Next to each pot is a gray docking station for the robots.

Each plant and robot is connected to a computer network. The plants, through sensors in their soil, can tell the network when they need water or fertilizer, while the robots use a camera to inventory the plants’ fruit. The robots also are programmed with a rudimentary growth model of the cherry tomato plants, which tells them roughly when a tomato will be ripe enough to be picked.

But the students quickly encountered challenges, both robotic and biologic.

Huan Liu, a 21-year-old computer science major, said designing the robot to pick the delicate tomatoes was made more difficult because the fruit would grow in unreachable places, such as behind stems or above where the robot’s arm could reach.

“The tomatoes, they come out of nowhere, or just in weird places,” Liu said.