SDSU robotics professor’s research explores topics from Apple to apples

Shabbir Ahmed, an assistant professor of artificial intelligence and robotics in South Dakota State University’s Department of Mechanical Engineering, is working on two projects funded by the National Science Foundation.

The first NSF grant, related to the Apple devices you may own, led to a recent publication in a prestigious journal, which proposed Koopman-based modeling approach in lithium-ion batteries.

His newest NSF grant has more to do with the fruit that may be in your kitchen.

Modeling battery degradation

For the past two years, a $200,000 grant from the NSF has funded Ahmed’s research on lithium-ion batteries. That’s the battery in your phone or laptop.

In September 2026, his research was published in the Journal of Energy Storage, a well-regarded journal in the field. He sought to address an important — and common — problem.

“As you use your phone, you will notice that, after one or two years, your phone cannot hold an appropriate amount of charge. That means the battery is degrading gradually,” he said.

Ahmed’s lab studied how the degradation occurs and built a model to predict the degradation of a particular battery. It’s not a graph with a straight line pointing down, Ahmed said. It’s a complicated, nonlinear system.

If the battery of a phone is out of charge, it’s easy to recharge as long as there’s an outlet and charger nearby. That isn’t always the case for other products powered by lithium-ion batteries, like drones, electric vehicles or electric aircraft.

“What about when you’re flying or driving long distances? You may need to charge it instantly, but you don’t have the facility or luxury at that moment,” Ahmed said. “So, it’s very important to know beforehand how much charge your battery can hold and how much degradation has occurred.”

The number on a display screen showing the battery’s percentage can’t always be trusted.

“You check your battery. How much charge is there? Let’s say it is 50%. How accurate is that? You believe whatever is given by your phone screen, but the algorithms behind it are approximating the estimate. It’s not exactly 50%,” he said.

It could be 48% or 45%. For a mobile phone, that doesn’t make a meaningful difference. For a drone or electric car traveling a long distance, it can affect whether the vehicle arrives at its destination.

Ahmed’s model is freely available to anyone and is available in the journal.

“Let’s say you are from Apple or Tesla or from any other company, and the researchers in that company can go through our paper, pick up the algorithm and the model, and they can apply this algorithm in their own case,” he said.

Since the project was funded by the NSF, all fruits from the research are shared freely for the public benefit.

Building robotic apple pickers

Ahmed’s new project, supported by $1 million in NSF funds over four years, aims to teach machines to see hidden fruit in agricultural fields and orchards.

This research is done in conjunction with a researcher at the University of California, Los Angeles, who works in computation and machine learning. Ahmed will do the experimentation at SDSU.

“There are a lot of large apple orchards, but not enough people to pick them. The apples may get rotten if they are not picked in time,” Ahmed said.

What if there was an automated, fruit-picking robot that could pick those fruits before they go bad?

There’s one problem: The robot needs to see all the apples in the tree, even the ones hidden by leaves and other branches.

“Your mobile camera is an RGB one, like red, green and blue. It can only detect those colors. If an apple is behind the leaves, this type of camera cannot see it. The leaves occlude, or hide, the fruits from view,” Ahmed explained.

Ahmed and his partner decided to try thermal cameras, which can recognize apples from leaves as well as apples hidden behind leaves. They all give off different heat signatures and are picked up differently by the camera.

At the University of California, Los Angeles, Ahmed’s partner will work on the robot’s machine learning algorithm. Ahmed and his lab are working on the hardware side.

But first, they’ll need to bring in an apple orchard to test the model on.

“We’ll create demo trees with demo fruits and demo leaves. We can also probably cut a branch of an apple tree and bring it into my lab. At the end of this, once we have perfected our model, then we test it in an open field,” he said

Since South Dakota is an agricultural state, there will be plenty of places to put his robotics to the test.