Table 21 Deviation errors on circular and square paths in full workspace.

From: Path-based evaluation of deep learning models for solving inverse kinematics in a revolute-prismatic robot

Model

Path shape

Deviation error (mm)

DFNN with K-fold CV (single o/p)

Square

1.594

DFNN with K-fold CV (dual o/p)

4.861

DFNN without CV

8.41

LSTM

5.775

GRU

24.13

DFNN with K-fold CV (single o/p)

Circle

2.084

DFNN with K-fold CV (dual o/p)

1.907

DFNN without CV

5.214

LSTM

0.873

GRU

4.076

Back to article page