Abstract: A difficult problem of picking robot visual position in the natural environment is the disturbance because of interference factors. The angular size of disturbed fruit was calculated to determine the three states of fruit that static, micro disturbance and large disturbance; For static and micro disturbance litchi, the improved FCM fuzzy clustering method was used to segment litchi image to acquire litchi fruit and fruit stalk, then the Hough linear fitting was used to determine the effective picking area and picking point on the fruit stalk, so realizing the visual space positioning of litchi picking point, and the simulation of manipulator picking disturbance target was carried on using the virtual reality technology. The visual positioning test results show that the visual positioning depth error values of static and micro disturbance litchis picking points are up to 2.3 cm, and the smallest is 0.6 cm, which shows the rationality and effectiveness of this method.

Key words: picking robot, dynamic target, lichi disturbance, vision localization, system simulation