DolbotX — National Defense Robot
GitHub: Highsky7/DolbotX Award: 🥈 Hanwha National Defense Award — 2nd Place, 2025 Army Chief of Staff Cup
Overview
DolbotX is an autonomous ground robot developed for the 2025 Army Chief of Staff Cup National Defense Robot Competition. As Software Team Lead, I was responsible for the full perception-to-control software stack, from training the vision models to integrating the ROS2 pipeline on the physical robot.
Key Contributions
Perception — Custom YOLOv8 Object Detection
- Collected and annotated a domain-specific dataset for competition obstacles and targets
- Fine-tuned a YOLOv8 model achieving robust real-time detection under varying lighting conditions
- Developed a driving area segmentation module to determine safe traversal regions
Optimization — ONNX Export & Deployment
- Exported trained models to ONNX format for hardware-accelerated inference on the embedded compute unit
- Achieved target latency for real-time closed-loop operation
Control — ROS2 Manipulator Pipeline
- Built the complete ROS2-based software stack covering perception, task planning, and manipulator control
- Integrated perception outputs with a finite-state machine for autonomous task execution
Technical Stack
| Category | Technologies |
|---|---|
| Vision | YOLOv8 (Ultralytics), Custom segmentation |
| Optimization | ONNX Runtime |
| Robotics | ROS2, Python |
| ML Framework | PyTorch |
| Hardware | Embedded compute (on-board) |
Results
- Hanwha National Defense Award — 2nd Place at the 2025 Army Chief of Staff Cup National Defense Robot Competition