Robotic Cameras Learn To Follow Basketball

Courtside cameras could be computer-controlled sooner than you think

2 min read
Basketball game between the Chicago Bulls and the Utah Jazz
Basketball games move fast, but new algorithms could help robotic cameras keep up with the action.
Charles Rex Arbogast/AP Photo

We’re getting used to the idea of robots taking our jobs in fields like manufacturing, but should a courtside cameraman for the NBA be worried? Until recently, that seemed like a safe gig to stay in human hands. But last weekened, Disney Research scientists reported that they’ve made strides in teaching automated cameras to track the action of a basketball game the same way a human camera operator does. 

While AI-operated cameras are not exactly groundbreaking, there’s a lot of room for improvement in the systems. That’s especially the case when it comes to fast-paced events like basketball games. Training a robotic camera to simply follow the ball results in jerky camera motions that can make a game hard to watch. That’s a concern for Disney, which owns ESPN and has a vested interest in presenting live sporting events at their finest.

At Disney Research, engineer Peter Carr and PhD student Jianhui Chen aimed to teach robotic cameras how to follow the game more like professional camera operators by anticipating where the ball is going to be rather than trying to keep an eye on where it is. 

"We don't use any direct information about the ball's location because tracking the ball with a single camera is difficult," Carr said in a statement. "But players are coached to be in the right place at the right time, so their formations usually give strong clues about the ball's location."

Carr and Chen’s system breaks the basketball court down into quadrants, and tracks the motion in those quadrants to create a map of where players are. That map provides the data an automated camera needs to predict where those players will be next, letting the camera setup shots of the action on the court instead of trying to react to it.

The algorithm can’t appreciate the grace of a pull-up jumper or the teamwork that goes into a well-executed pick and roll—yet. It is getting better at identifying what’s likely to happen next in a game, though. In a test conducted at a high school basketball game, though, stationary robotic camera closely aped the pans, tilts and zooms of a human operator filming the same game. Carr and Chen presented their first findings over the weekend at the IEEE Winter Conference on Applications of Computer Vision, and you can take a look at the results here.

The algorithm has only been tested on basketball so far, but Carr and Chen hope that it can be adapted to other sports as it is developed further. So someday soon, computers may not just be better at playing games like poker and beer pong than we are. They may even be better at watching them. 

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman

This article is part of our special report on AI, “The Great AI Reckoning.

"I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less