Work in Progress — Active Development
IRIS is an embodied robotics project focused on integrating perception, manipulation, planning, and intelligent autonomy into a unified robotic system.
IRIS is a full-stack robotics system being developed around:
- 3D-printed 6-DOF robotic arm for manipulation
- Multi-camera perception pipeline for tracking and scene understanding
- ROS 2 + MoveIt for robot control and motion planning
- Classical + AI vision methods including OpenCV, YOLO, and optical flow
- Agentic planning concepts for higher-level autonomous task execution
The long-term goal is to build an embodied robotic system capable of perception-driven interaction and intelligent task execution.
This project is under active development.
Current work includes:
- Robotic arm integration
- Dual-camera perception pipeline
- Object tracking and world-state modeling
- ROS communication architecture
- Motion planning experiments
- Agentic control exploration
Documentation, architecture, and code structure may change frequently.
IRIS
├── Perception
│ ├── Multi-camera vision
│ ├── Object detection/tracking
│ └── World modeling
│
├── Manipulation
│ ├── 6-DOF robotic arm
│ ├── Motion planning
│ └── Grasp execution
│
├── Control
│ ├── ROS 2 interfaces
│ ├── MoveIt integration
│ └── Perception-to-action pipeline
│
└── Intelligence
├── Agentic planning
├── Decision making
└── Autonomous task execution
- Python
- C++
- ROS 2
- MoveIt
- OpenCV
- YOLO
- Core robotic arm control
- Dual-camera perception integration
- Tracking + world-state representation
- Vision-guided manipulation
- Agent-based task planning
- Autonomous manipulation demos
This repository currently serves as:
- Development workspace
- Research/prototyping environment
- System architecture repository
- Ongoing documentation of project evolution
More structured documentation and demos will be added as the project matures.
IRIS is currently an experimental long-term robotics project under active iteration.
Updates, demos, and expanded documentation will be added progressively.