The Application of Computer Aided Design on Robotic Path Planning and Control

• Main Authors: Chen, Hsu Min, 陳旭敏
• Other Authors: Chang, Yih Feng ; Lin, Shyng Her
• Format: Others
• Language: zh-TW
• Published: 1994
• Online Access: http://ndltd.ncl.edu.tw/handle/95614744061385485998

碩士 === 中原大學 === 機械工程研究所 === 82 === Most of the robotic path plannings rely on users to guide the robot to move along the working path , or to plan the trace by a teaching pendant while robot recording the trajectory . These human teaching methods are applicable to the case of large number production in a unaltered manufacturing process . When the tasks are altered constantly as usually occured in flexible manufacturing system , this routine will become unbearable and inefficient. This report employs the highly used drawing software - AutoCAD , to develop a robot path planning tool . At present, AutoCAD is the most popular drawing tool on the personal computer , it is very convenient , fully developed , and easy to use . So , if applying these capacities on the robotic trajectory planning and motion control , it will increase path programming efficiency and provide a convenient , functional and economical man-machine interface to users . The objective is to apply the AutoCAD as the interface between users and robot . Users need only using a mouse to draw a working path on the screen and keying in the desired robot speed , a developed program will automatically convert the drawing into trajectory commands and feed them into the working robot . The actual trace of the robot will also be fed back to the program and shown on the screen as a drawing added on the origional planned path . Users could immediately observe the results and decide to accept the response or not, so as to change the robot moving speed if needed . The experimental instrument in this work is a rectangular robot driven by stepping motors . Due to certain limitation of the control hardware and the open-loop control nature of the robot , the motion precision is degraded during tracking control . To improve this deficiency , this report trains a set of neural networks to predict the tracking error of a given task before execution ,then compensates the control commands by adjusting the planned path to increase tracking precision .