Font Size: a A A

Statistical and deterministic models for robot inaccuracy compensation using relative feature recognition

Posted on:1989-04-21Degree:Ph.DType:Dissertation
University:Brigham Young UniversityCandidate:Davies, Brady RFull Text:PDF
GTID:1478390017955445Subject:Engineering
Abstract/Summary:
Currently, positionally accurate industrial robot tasks are programmed using a teach pendant. Using the teach pendant, the robot is literally led-through or taught each position comprising the robot task.; Robots can be programmed more efficiently off-line, since off-line programming can include task simulation and high level task definition. Robot inaccuracy severely limits the practicality of off-line programming since paths planned off-line generally deviate significantly from desired paths.; Using robot inaccuracy compensation, the accuracy of paths planned off-line were significantly improved. Robot inaccuracy was modeled with both kinematic and stochastic models. These models were calibrated by measuring robot repeatability and robot inaccuracy relative to parts being processed by the robot. The calibration process was automated by attaching vision and probing sensors to the robot arm.; This approach reduced inaccuracy of a GMF S-200 robot from 3.5 mm to less than 0.86 mm. Additionally, repeatability error of this GMF S-200 robot was reduced by 39%.
Keywords/Search Tags:Robot inaccuracy, GMF S-200 robot, Teach pendant, Models, Paths planned off-line
Related items