Home > Research > Table-tennis robot

Analysis of human skills and transfer to a robot

  An excellent table tennis player appropriately reacts to the constantly changing state of the ball.

  Because the way of hitting the ball may change according to the elements that influence the ball's flight characteristics (speed, direction, orbit, spin, rebound characteristics, etc.) or the position of the opponent, to be exact, all actions differ from those that were taken before.

  We have analyzed the skill of human's table-tennis movement.

  Movement patterns of the body and the timing to generate them involved in swinging is extracted from human movements observation with PCA and generalized as the schema of GMP. Transferring these skills to robots, we could make the robot properly play table-tennis with a human.



Research introduction