The problem and project leadership
Funded under the TÜBİTAK 2209/A University Students Research Projects Support Program, this initiative investigated an autonomous mobile robot performing line-following, QR dynamic task interpretation, computer vision object/shape recognition, and six-axis robotic arm grasping in a single unified workflow.
As Project Director, I led system architecture, electronic integration, and cross-subsystem execution. The goal was to establish a verifiable, state-machine-driven task pipeline linking computer vision with real-time embedded motor and arm actuation.
System architecture
A Raspberry Pi 5 handled the high-level workflow and computer-vision processing, while an Arduino UNO managed time-sensitive motion and hardware control. The two control layers communicated over UART.
The Python application used OpenCV for vision processing, Pyzbar for QR decoding, Pyserial for serial communication, and Flask for a local observation surface. Hardware control was implemented in Arduino C++.
- Camera and QROpenCV + Pyzbar
- Raspberry Pi 5Python state machine
- Arduino UNOUART + Arduino C++
- Motion and armMotors + six axes
State machine and task flow
Robot behaviour was modelled as a state machine with explicit transitions. This made it possible to see when line following, QR task intake, perception, arm movement, and recovery steps should begin and end.
Perception and motion components
- Route traversal through line following
- Pyzbar-based QR task decoding
- Object and shape detection with OpenCV
- Physical task execution with a six-axis robotic arm
- UART messaging between the Raspberry Pi 5 and Arduino UNO
- Controlled stop and retry paths for unexpected states
Test approach and verified metrics
The prototype was evaluated over 50 test rounds. Full-chain success measured 64%, while individual subtasks measured between 76% and 88%.
These figures describe a controlled research prototype, not production-grade reliability. The difference between subtask and full-chain results directed the next iteration toward transitions between stages and recovery behaviour.
- Test rounds
- 50
- Controlled prototype conditions
- Full task chain
- 64%
- End-to-end completed rounds
- Subtasks
- 76–88%
- Measured range by task
Research outcome and academic paper
The project delivered a verified engineering framework proving how computer vision and embedded hardware control can be integrated as an end-to-end task loop under quantifiable metrics.
Derived directly from these technical results, the academic paper titled “Development of an Autonomous Robot with Vision-Based Lane Tracking, QR Code Reading, and Predefined Servo Manipulation Grasping” was submitted to the Journal of Computer Sciences and Engineering. The research specifically addresses low-cost autonomous mobile robotics, computer vision, dual-controller topologies, and task-based robotic automation.