PingPongRobot.cn

AI Table Tennis Rally Robot System

Designed for indoor entertainment, training and science engagement. The system tracks the ball, predicts its bounce and strike position, and controls the paddle to make an intelligent return.

AI Table Tennis Rally Robot System

A millisecond closed loop from perception to return

Sensing, prediction, planning, servo execution and 3D visualization cooperate over a real-time industrial control network.

01

Stereo high-speed vision

In-house stable tracking at 100–200 FPS.

02

Trajectory prediction

Up to 250 prediction updates per second.

03

Linear parallel robot

Approx. 2.5 m/s end speed; full travel within 0.40 s.

04

2-DOF servo paddle

About 0.12 s for a 90° action, with programmable strategies.

View specifications

Stereo high-speed vision
100–200 FPS
Trajectory prediction
≤ 250/s
Linear parallel robot
≈ 2.5 m/s · ≤ 0.40 s
2-DOF servo paddle
90° · ≈ 0.12 s
75%+ official algorithm return rate
75%+ · ≈ 3.5 m/s
Core technology and architecture
A real-time industrial network, CAN servo control, stereo vision, 3D reconstruction, online prediction and a lightweight parallel actuator form one closed loop.
Stay open
Vision, actuator and 3D simulation interfaces support secondary development and OTA updates.
Pmax
3 kW
m
≈ 350 kg

Physical robot and digital twin

2-DOF servo paddle

2-DOF servo paddle

About 0.12 s for a 90° action, with programmable strategies.

Stereo high-speed vision

Stereo high-speed vision

In-house stable tracking at 100–200 FPS.