Robotic desk tennis system predicts ball trajectory and adapts swing in actual time

Over the previous a long time, roboticists have launched numerous programs that may replicate particular human motions and behaviors with exceptional accuracy. A few of these robots may even compete with different robots or people in particular sports activities, such because the robots showcased on the RoboCup, a global robotics occasion at which robots play soccer with one another.

Researchers on the Massachusetts Institute of Expertise (MIT)’s Biomimetic Robotics Laboratory not too long ago launched a brand new robotic desk tennis platform that may efficiently and quickly hit balls with a desk tennis racket. Their platform, outlined in a paper printed on the arXiv preprint server, can reproduce numerous desk tennis hit kinds and spin the ball in numerous instructions with excessive precision.

“The Biomimetic Robotics Lab at MIT has all the time strived to create performant robotic programs by innovating on each the {hardware} and management as seen with the Mini-Cheetah,” Kendrick Cancio, co-author of the paper, informed Tech Xplore.

“We had a chance to create this desk tennis system on behalf of the Robotics and AI Institute as a platform to discover dynamic manipulation with the final word aim of reaching human parity at table-tennis on a dynamic anthropomorphic platform.”

The Biomimetic Robotics Lab conducts analysis specializing in two major areas of robotic management: dynamic-legged locomotion (i.e., the versatile motion of legged robots) and the fast manipulation of objects. These two areas of robotics analysis include their very own distinctive challenges.

Relating to legged robotic locomotion, a key problem is successfully dealing with disturbances within the setting, whereas profitable object manipulation entails precisely performing desired maneuvers.

“Desk tennis blurs the strains between these management points with adaptiveness and precision required as extra details about an incoming ball turns into obtainable,” David Nguyen, co-author of the paper, informed Tech Xplore. “This makes it a really distinctive management downside that we expect we will hit out of the park with our customized robotics {hardware}.”

The brand new platform developed by Nguyen, Cancio and Sangbae Kim consists of a robotic arm and a management algorithm. The algorithm can predict the trail of an incoming ball and plan the actions that the arm ought to carry out when swinging the racket to hit the ball, whereas additionally assembly specified strike situations.

“Even whereas swinging, the trail the arm takes is dynamically up to date to make sure the paddle reaches the ball on the proper location, pace, and orientation,” defined Nguyen.

“We discovered that planning for your entire swing quite than simply the long run actions was extra dependable however requires extra aggressive maneuvers from our arm. This can be a distinctive benefit of our work since we will push the boundaries of our customized {hardware} excess of an off-the-shelf system.”

The robotic programs developed by the researchers has two major parts, known as the notion and actuation modules. The notion module consists of an off-the-shelf movement monitoring system, which might localize the customized desk tennis balls developed by the researchers.

“We predict the ball’s trajectory to get an anticipated strike location and strike time,” defined Cancio.

“On the similar time, we now have a nonlinear optimization downside which makes use of these values together with info on how we want to strike the ball to generate a swing trajectory for the arm. Our mannequin predictive controller frequently solves for this arm trajectory and permits the arm to react as we get up to date positions of the ball.”

The robotic arm built-in within the staff’s desk tennis platform is a personalized model of a humanoid arm developed at MIT. This arm has high-torque and low rotor inertia, two traits that permit it not solely to swing shortly, but in addition to react shortly and adapt its trajectories if preliminary predictions a couple of ball’s trajectory are fallacious.

“We reveal a capability to adapt a trajectory to exactly meet a dynamically shifting object,” mentioned Nguyen. “Though desk tennis will not be going to save lots of any lives, this type of management could possibly be utilized in tough search and rescue conditions the place a extra basic robotic like a humanoid would possibly must intercept an object.”

Nguyen, Cancio and Kim evaluated their robotic desk tennis platform in a collection of real-world experiments and located that it carried out remarkably effectively. In these preliminary assessments, the robotic arm hit incoming balls with successful price of 88% and a mean exit velocity of 11 m/s, following three distinct hitting kinds.

“Proper now, robotics is usually break up between model-based approaches and reinforcement studying approaches, with some anticipating the latter to be the catch-all device within the close to future,” mentioned Cancio. “We present that constraint-based optimization nonetheless has a spot for performant programs and hope to leverage the advantages of every when acceptable.”

The brand new system developed by this staff of researchers might quickly encourage different roboticists to develop related automated desk tennis platforms. As well as, Nguyen, Cancio and Kim hope to use the {hardware} and management algorithm they developed to different dynamic manipulation duties.

“Since we submitted the paper again in September, we now have completed loads to enhance the capabilities of the system,” added Nguyen. “Particularly, we are actually capable of intention at particular places on the desk and plan for your entire trajectory and call between the ball and paddle.”

As a part of their future research, the researchers plan to additional improve the desk tennis robotic platform’s capabilities. By broadening the MIT humanoid arm’s workspace utilizing a gantry (i.e., a construction supporting the arm), as an illustration, they might permit it to play whole video games of desk tennis in opposition to human customers.

“We intention to maintain pushing the efficiency of our system by increasing the workspace utilizing a gantry and considerably rising our ball exit velocities,” added Cancio.

“We’d additionally like to maneuver in the direction of monitoring commonplace desk tennis balls to make higher comparisons in opposition to people and different robotic programs alike.”

© 2025 Science X Community