A learning approach to robotic table tennis

Matsushima, Michiya; Hashimoto, Tadashi; Takeuchi, Masahiro; Miyazaki, Fumio

doi:10.1109/tro.2005.844689

Cited by 120 publications

(70 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Note that our results differ significantly from previous approaches as we use a framework that allows us to learn striking movements from human demonstrations unlike previous work in batting [Senoo et al, 2006] and table tennis [Andersson, 1988]. Unlike baseball which only requires four degrees of freedom (as, e.g., in [Senoo et al, 2006] who used a 4 DoF WAM arm in a manually coded high speed setting), and previous work in table tennis (which had only low-inertia, was overpowered and had mostly prismatic joints [Andersson, 1988, Fässler et al, 1990, Matsushima et al, 2005), we use a full seven degrees of freedom revolutionary joint robot and, thus, have to deal with larger inertia as the wrist adds roughly 2.5k g weight at the elbow. Hence, it was essential to train trajectories by imitation learning that distribute the torques well over the redundant joints as the human teacher was suffering from the same constraints.…”

Section: Playing Against a Ball Launchermentioning

confidence: 99%

Learning motor skills: from algorithms to robot experiments

Kober

Peters

2014

It - Information Technology

View full text Add to dashboard Cite

Die Veröffentlichung steht unter folgender Creative Commons Lizenz: Namensnennung -Keine kommerzielle Nutzung -Keine Bearbeitung 2.0 Deutschland http://creativecommons.org/licenses/by-nc-nd/2.0/de/ Abstract Ever since the word "robot" was introduced to the English language by KarelČapek's play "Rossum's Universal Robots" in 1921, robots have been expected to become part of our daily lives. In recent years, robots such as autonomous vacuum cleaners, lawn mowers, and window cleaners, as well as a huge number of toys have been made commercially available. However, a lot of additional research is required to turn robots into versatile household helpers and companions. One of the many challenges is that robots are still very specialized and cannot easily adapt to changing environments and requirements. Since the 1960s, scientists attempt to provide robots with more autonomy, adaptability, and intelligence. Research in this field is still very active but has shifted focus from reasoning based methods towards statistical machine learning. Both navigation (i.e., moving in unknown or changing environments) and motor control (i.e., coordinating movements to perform skilled actions) are important sub-tasks.In this thesis, we will discuss approaches that allow robots to learn motor skills. We mainly consider tasks that need to take into account the dynamic behavior of the robot and its environment, where a kinematic movement plan is not sufficient. The presented tasks correspond to sports and games but the presented techniques will also be applicable to more mundane household tasks. Motor skills can often be represented by motor primitives. Such motor primitives encode elemental motions which can be generalized, sequenced, and combined to achieve more complex tasks. For example, a forehand and a backhand could be seen as two different motor primitives of playing table tennis. We show how motor primitives can be employed to learn motor skills on three different levels. First, we discuss how a single motor skill, represented by a motor primitive, can be learned using reinforcement learning. Second, we show how such learned motor primitives can be generalized to new situations. Finally, we present first steps towards using motor primitives in a hierarchical setting and how several motor primitives can be combined to achieve more complex tasks.To date, there have been a number of successful applications of learning motor primitives employing imitation learning. However, many interesting motor learning problems are high-dimensional reinforcement learning problems which are often beyond the reach of current reinforcement learning methods. We review research on reinforcement learning applied to robotics and point out key challenges and important strategies to render reinforcement learning tractable. Based on these insights, we introduce novel learning approaches both for single and generalized motor skills.For learning single motor skills, we study parametrized policy search methods and introduce a framework of reward-weighted imi...

show abstract

Section: Playing Against a Ball Launchermentioning

confidence: 99%

Learning motor skills: from algorithms to robot experiments

Kober

Peters

2014

It - Information Technology

View full text Add to dashboard Cite

show abstract

“…For a robot to complete this task, the trajectory of a moving ball must be predicted in advance. We use the physicalbased method rather than the regression one [2,24] or the quadratic fitting one [1] to predict ball's trajectory.…”

Section: Trajectory Predictionmentioning

confidence: 99%

“…The constraint of detecting region and the simple five-point ball representation are the main reasons for the fast-processing performance. Unlike some other vision systems [2,10,12] that rely on color information to detect a ball, our system mainly use the motion feature and our five-point feature to detect a ball. We consider that color feature is not robust enough since a ball shows different appearances under different illumination conditions.…”

Section: Comparisons With Previous Vision Systemsmentioning

confidence: 99%

“…Andersson [1] firstly constructed a ping-pong robot capable of playing against humans and machines. The ping-pong robot designed by Miyazaki et al can play against humans [2] or play against wall [3]. The ping-pong system described in [4,5] is capable of returning an incoming ball to the opponent's court.…”

Section: Introductionmentioning

confidence: 99%

“…But synchronization may become difficult when they capture images in high speed. In [2], the stereo vision system named Quick MAG III is used to extract a ball every 1/60 s. However, the system excessively relies on ball's color. The vision system developed in [10] only needs a single CCD video camera, based on detecting a ball and the shadow it projects on the table.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Short-Baseline Binocular Vision System for a Humanoid Ping-Pong Robot

Tian

Sun

Tang

2011

J Intell Robot Syst

View full text Add to dashboard Cite

We develop a short-baseline vision system for a humanoid ping-pong robot. The vision system can provide four-dimensional space-time information and can predict the future trajectory of a ball. Short baseline poses special challenges for achieving sufficient 3-D reconstruction and prediction accuracy within limited processing time. We propose two algorithms including direct calibration of projection matrix and Gaussian-fitting based ball-center location to guarantee the 3-D reconstruction accuracy; we propose algorithm of five-point based ball representation and utilize the constraint of ball detecting region to guarantee the processing speed; we also propose algorithm of smoothing-based trajectory prediction to improve the prediction accuracy. Experimental results show the accuracy and the speed of our vision system can meet the requirements of a humanoid ping-pong robot.

show abstract

A Learning Model for Racket Motion Decision in Ping‐Pong Robotic System

Guodong

et al. 2014

Asian Journal of Control

View full text Add to dashboard Cite

In this paper, a model of computing the desired velocity of returning balls is developed so that the robot could return the incoming ball to a desired point on the table with specified landing velocity. At first, three polynomials of the flying time are applied to fit the ball's trajectory while neglecting the effort of Magnus force; one result of returning velocity is obtained after the coefficients of the polynomials are estimated using the LM (Levenberg‐Marquardt) algorithm. Then a new configuration based on area segmentation is proposed which is used in storing and replacing experimental data; another result of the velocity is received by fitting the stored data. The weighted average of the two above mentioned results is regarded as the original value of the fuzzy correcting algorithm. With the analysis of the influence of the desired velocity on the errors between the actual landing point and the desired point, the fuzzy rule is designed. The eventual returning velocity of the ball is obtained using the fuzzy correcting algorithm based on the errors of the landing point predicted by the flying model. Finally, experiments are conducted to verify the performance of the proposed method.

show abstract

A learning approach to robotic table tennis

Cited by 120 publications

References 6 publications

Learning motor skills: from algorithms to robot experiments

Learning motor skills: from algorithms to robot experiments

Short-Baseline Binocular Vision System for a Humanoid Ping-Pong Robot

A Learning Model for Racket Motion Decision in Ping‐Pong Robotic System

Contact Info

Product

Resources

About