Abstract: To address the issue of frequent manual ball picking in table tennis, an intelligent table tennis ball-picking robot is designed. The hardware mainly includes 3 systems: ball-collection, ball-storage, and ball-ejecting. The coordination of the 3 systems implements the functions of collecting contacted balls into the storage system and making the storage system empty once the capacity is reached. The software implements the functions of planning suitable paths and controlling the robot to approach the balls, including a direction correction algorithm, a data fusion algorithm, and a path planning algorithm, which are based on nonlinear PID, Kalman filter, and genetic algorithm respectively. To verify the robot effectiveness, 4 experiments are designed: real-time test, global positioning system error test, nonlinear PID simulation experiment, and physical test. The results show that the single-run execution time of the proposed algorithm does not exceed 5 μs, with an average error of approximately 12.05 cm. The physical prototype achieves a 100% success rate in ball collection, and its average speed reaches 82.5% of the specified baseline velocity, demonstrating the feasibility of the design scheme.