alan yin
← Work

NARX Neural-Network Robotic Arm Control

[One or two sentences: what the controller did, and what the hardware was for.]

Year
2024 – 2025
Status
Published — Springer, 2025
Context
Research internship, Department of Aerospace Engineering, Toronto Metropolitan University
Role
Control implementation, test hardware, co-author
Stack
MATLAB · Python · SolidWorks
Publication
Springer Lecture Notes in Electrical Engineering, Vol. 1468
CAD front view of the robotic actuator in its test frame: a belt-driven two-stage joint mounted between two perforated aluminium uprights, with the drive pulley below and the output stage above.
[The actuator in its test frame — name what the reader is looking at.]

What it does

[What the controller does — what it learns, and what it sends to the arm.]

[Why a NARX network rather than a conventional controller.]

How it works

[The mechanism: inputs, outputs, training data, and the path from network to motor.]

The test hardware

[Why the arm needed a fixture of its own.]

[What the fixture measured, and what counted as agreement between commanded and measured motion.]

[What the clip shows — it is short, so say what to watch for.]

What I built

  • [The control implementation.]
  • [The validation hardware.]
  • [Your part in the paper and the presentation.]

Publication and presentation

Paper“NARX Neural Network–Based Control of Robotic Arm”
VenueSpringer Lecture Notes in Electrical Engineering, Vol. 1468 (2025)
PresentedIEEE IEMTRONICS 2025 — Best Presenter Award

[Optional: anything worth saying about the conference, or delete this paragraph.]

Limitations and what comes next

  • [A real limitation.]
  • [What you would do differently.]