Design and evaluation of an embedded automation system for optimized cut-shape placement on coconut shells in...

This paper presents the design and evaluation of a novel embedded automation system for optimized cut-shape placement on coconut shells in sustainable key tag manufacturing. The system addresses inefficiencies in manual marking by improving speed, accuracy, and material utilization, while ensuring affordability and suitability for decentralized artisanal contexts.

The system integrates a Raspberry Pi 4, VL53L0X Time-of-Flight sensor, Pi-Camera, MG995 servos, and a low-power laser module, orchestrated through a Python-based finite-state machine (FSM). A lightweight CNN model deployed via TensorFlow Lite enables real-time classification of shell geometries. Comparative experiments were conducted on 20 coconut shells under both manual and automated marking conditions. Evaluation metrics included marking time, accuracy of placement, material utilization, and productive yield.

The proposed system demonstrated significant improvements over manual marking: marking time was reduced by 26.0%, placement accuracy improved by 79.4%, and material utilization increased by 38.4%. The average number of usable cuts per shell increased from 2.85 (manual) to 3.89 (automated), representing a 36.5% gain in productive yield. All improvements were statistically significant (p < 0.05).

This is the first known application of a low-cost, edge-AI-enabled embedded system combining ToF sensing, CNN-based shape classification, and adaptive actuation for real-time marking on irregular natural materials. The system dynamically adapts to geometric variability in biodegradable substrates, an area underserved by conventional automation solutions.

The system is cost-effective (<$132), lightweight (<2.5 kg), and fully modular, making it suitable for deployment in SMEs, rural micro-factories, and vocational training centers.

By augmenting artisanal labor rather than replacing it, the system supports inclusive digital transformation in low-resource environments and contributes to SDGs 8, 9, and 12. It is also generalizable to other natural materials, supporting broader applications in sustainable, circular manufacturing.