Manufacturing Technology Today

Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Identification of the End-Effector Positioning Errors of a 2-Link Robot Arm Using Back Propagation Neural Network


Affiliations
1 Dept. of Mechanical Engg., V R Siddhartha Engg. College, Vijayawada, India
2 Dept. of Mechanical Engg., A U College of Engg., Visakhapatnam, India
     

   Subscribe/Renew Journal


In this work, a 2 link planar robot arm is studied for the inter relation between the inclinations of links and the end-effector position. Due to the deflection of the link under loading, the end-effector needs specific corrections to keep it at a desired location in space. The values are theoretically calculated for specific number of examples with varied loading and link inclinations. C-programme is composed to calculate the theoretical deflections and those results are cross checked with the results from ANSYS 10.0 analysis. These values are assigned as inputs to train a BPNN suitably. The characteristics of BPNN are varied to find the accuracy of the network in predicting the location corrections of end-effector. The network is trained with 30 sets of sample i.e. the network is adjusted for the function and weights according to the input and output values. After the formation of the function, the network is tested for different input values.
User
Subscription Login to verify subscription
Notifications
Font Size

Abstract Views: 223

PDF Views: 0




  • Identification of the End-Effector Positioning Errors of a 2-Link Robot Arm Using Back Propagation Neural Network

Abstract Views: 223  |  PDF Views: 0

Authors

A. Mallikarjuna Rao
Dept. of Mechanical Engg., V R Siddhartha Engg. College, Vijayawada, India
G. Dilli Babu
Dept. of Mechanical Engg., V R Siddhartha Engg. College, Vijayawada, India
B. S. K. Sundara Siva Rao
Dept. of Mechanical Engg., A U College of Engg., Visakhapatnam, India

Abstract


In this work, a 2 link planar robot arm is studied for the inter relation between the inclinations of links and the end-effector position. Due to the deflection of the link under loading, the end-effector needs specific corrections to keep it at a desired location in space. The values are theoretically calculated for specific number of examples with varied loading and link inclinations. C-programme is composed to calculate the theoretical deflections and those results are cross checked with the results from ANSYS 10.0 analysis. These values are assigned as inputs to train a BPNN suitably. The characteristics of BPNN are varied to find the accuracy of the network in predicting the location corrections of end-effector. The network is trained with 30 sets of sample i.e. the network is adjusted for the function and weights according to the input and output values. After the formation of the function, the network is tested for different input values.