A New Model that Addresses the ‘Outfielder Problem’

Andrew Corselli

How does a tennis player like Carlos Alcaraz decide where to run to return Novak Djokovic’s ball by just looking at the ball’s initial position? These behaviors, so common in elite athletes, are difficult to explain with current computational models, which assume that the players must continuously follow the ball with their eyes. Now, researchers at the University of Barcelona have developed a model that, by combining optical variables with environmental factors such as gravity, accurately predicts how a person will move to catch a moving object just from an initial glance. These results, published in the journal Royal Society Open Science, could have potential applications in fields such as robotics, sports training, or even space exploration.

The paper addresses the outfielder problem, which refers to the baseball player who stands in the outfield to catch the ball after it is hit. It is a classic challenge in physics and the neuroscience of movement, used to explore how humans and animals predict movements in a dynamic environment and how automated systems can be designed to mimic them.

Joan López-Moliner, professor at the UB’s Faculty of Psychology and member of the Institute of Neurosciences (UBneuro), has led the research and affirms that “faced with this problem, current models are based on guiding locomotion by continuously looking at the ball, while normally the elite athlete can run toward the ball without looking at it. Moreover, these models do not allow predictions of where the ball will go regarding the observer.”

The initial study was part of the doctoral thesis conducted by Borja Aguado, Co-Author and former member of the group, who, after a stay in Darmstadt (Germany), is now a researcher at the University of Vic.

The model integrates prior knowledge of the ball’s gravity and physical size into the visual information received in real time. “The model provides live signals that indicate the predicted position of the ball’s fall and the time remaining until it arrives, considering different gravity conditions. This makes it possible to predict precisely how a player will move to catch it, from the very beginning of the flight,” said López-Moliner, who also coordinates the Vision and Control of Action research group.

Despite the importance of gravity in anticipating trajectories, this is the first time this factor has been included in such a model. “This omission has overlooked the substantial influence that gravity exerts on the trajectory, which reflects a gap in the way existing models take into account environmental constants,” he said.

Moreover, the previous models cannot explain why humans perceive whether a ball is within reach or not to decide whether to start running. “Our model does account for this, as it indicates where the object will go regarding the player,” said the researcher.

Here is an exclusive Tech Briefs interview, edited for length and clarity, with López-Moliner.

Joan López-Moliner (Image: University of Barcelona)

Tech Briefs: What was the biggest technical challenge you faced while developing this predictive model?

López-Moliner: The model is just pure math, geometry. But we had to consider the perceptual uncertainty of the human visual system. So, the fact that we can show with equations that information is available and can be used to predict where a ball is going to be in the near future, doesn't mean that humans will use this information. There are some variables in the equations, we call optical variables, that are projections on the retina. So, some of them are perceived with a high uncertainty. Then we had to fit within the equations the uncertainty of the visual system in order to test whether they could use this information.

Tech Briefs: What was the catalyst for this project?

López-Moliner: One of my main lines of research is not just fielding, but all the models of what we call time-to-contact or time-to-collision estimates. So, how the visual system uses visual information in order to derive where some objects will be in the near future. It's a quite complex topic and has been done with simple trajectories, linear trajectories, but nothing with parabolic flights.

So, the main problem was standing on my desk for some time since we derived the mathematical equations. The main thing is that the previous theories don't like concepts like gravity or physical size. The dominant models want to rely only on optical variables; optical information is on the retina without any assumption about the state or context of the ball, like gravity, for example. So, we realized that when we introduced gravity, a known size, physical size, not angular size, which is the size on your retina, but the actual physical size of the ball, actual contextual variables, the equations became much simpler. So, estimating time to contact and final distance relative to the outfielder was easier than before — because before, there were very complex equations and no one believed that the brain could really solve them.

Tech Briefs: Do you have any updates you can share?

López-Moliner: Well, it's the very beginning. We are training neural networks based on reinforcement learning. It's more interactive. We give the agent different views of visual information, and it has to learn to be at the correct place at the right time. So, we are at the very beginning, trying different sizes of the neural networks, and basically the idea is that, ‘OK, the estimate of our model should be represented in the hidden layers of the neural network.’ We feed the network input information, the information that is on the retina, and, so far, we let the network find the internal model that allows it to predict where the ball is going to be.

So, we are at a very early stage. I have one student working on that, and I hope that by summer we may have some promising initial data.

Tech Briefs: Do you have any advice for researchers or engineers aiming to bring their ideas to fruition (broadly speaking)?

López-Moliner: Just be persistent. We had problems, difficulties in publishing the paper because the dominant theories don't like internal models that, for example, include gravity. So, we had to be really persistent and show the referees that the model could really account for the data, while the previous model couldn't. So, I would say that persistence in believing in the data. And then, persistence if they think they really have a model that can explain the data.