| Title |
Neural networks based robot motion improvement / |
| Translation of Title |
Roboto judesių gerinimas neuroniniais tinklais. |
| Authors |
Valaitis, Vytautas |
| Full Text |
|
| Pages |
93 |
| Keywords [eng] |
robotic motion ; inverse kinetatics ; neural networks ; perceptron. |
| Abstract [eng] |
This thesis focuses on nature-based algorithms to solve inverse kinematics and motion planning tasks of robotic systems and serial manipulators. Motions in nature can be classified to reflexes, partially and fully coordinated motions. Important step in motion execution – solving problem of inverse kinematics. Analytic method becomes insufficient in real world conditions. This research analyzes single layer and multi-layer perceptron learning in a changing task environment, and their learning rapidity. Methods to increase analytic algorithms accuracy while solving the inverse kinematics problem of a hexapod robot were introduced. Methods for trajectory planning using splines and primitives were analyzed. Algorithms were paralelized. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Doctoral thesis |
| Language |
English |
| Publication date |
2016 |
