Vision Options
Computer Vision - See the Field
Cameras let a robot detect game pieces, track targets, and figure out where it is on the field. AprilTags provide absolute field positioning, while object detection helps with game piece manipulation.
Without vision, your robot only knows where odometry says it is. With it, the robot can see the field and correct itself.
Why Vision Matters in FRC
Absolute Positioning
Game Piece Detection
Target Tracking
💡 Vision is Essential for Competitive Play
Modern FRC all but requires vision for accurate autonomous and teleop assists. Top teams run multiple cameras to keep more of the field in view.
FRC Vision System Options
Limelight
Dedicated vision hardware with integrated processing, LEDs, and a NetworkTables interface. Plug it in and it works.
✅ Advantages
- Zero code required for basic detection
- Built-in LED ring for consistent lighting
- Hardware-accelerated processing
- Web interface for tuning pipelines
- NetworkTables integration out-of-box
- Proven reliability in competition
⚠️ Limitations
- Higher cost (~$400-500)
- Proprietary hardware and software
- Limited customization vs open source
🎯 Best For
Teams that want proven vision hardware with minimal setup and are willing to pay for the convenience.
PhotonVision
Open-source vision software that runs on coprocessors (Raspberry Pi, Orange Pi, etc.). The cheaper, do-it-yourself route.
✅ Advantages
- Free and open source
- Works with any USB camera
- Active development and community
- Advanced AprilTag support
- Multi-tag pose estimation
- Lower total cost (~$100-150)
⚠️ Limitations
- Requires coprocessor setup
- More configuration complexity
- Camera and lighting selection matters
- Performance depends on hardware chosen
🎯 Best For
Teams that want cheap, flexible vision and don't mind the extra setup. Great for AprilTag localization.
Hardware Setup
🔌 Power & Connectivity
- Power: Connect Limelight to the PDH/PDP (12V). Do not use the VRM unless necessary.
- Ethernet: Connect directly to the radio or use a network switch if you have multiple devices.
- Mounting:Mount cameras where they can see the scoring tags while you're scoring, and not at the same height as the tags. The camera should view a tag at an angle (left or right, and up or down).
🏷️ AprilTag Configuration
Ensure your Limelight is up to date and configured for the current game's AprilTag field map.
- Update Limelight OS to the latest version via USB flash drive.
- Download the latest AprilTag Field Map from the Limelight Downloads page.
- Upload the map via the Limelight web interface (Hardware Manager 2.0).
Vision System Comparison
| Feature | Limelight | PhotonVision |
|---|---|---|
| Cost | $400-500 | $100-150 |
| Setup Complexity | Very Easy | Moderate |
| Hardware | Integrated unit | Coprocessor + Camera |
| AprilTag Support | Yes | Yes |
| Custom Pipelines | Web interface | Web interface + Code |
| Community Support | Large, established | Growing, active |
| Best Use Case | Plug-and-play reliability | Cost-effective flexibility |
AprilTag Localization
AprilTags are fiducial markers placed at known locations on the FRC field. Cameras can detect these tags and calculate the robot's absolute field position.
How AprilTags Work
- Tags have unique IDs corresponding to field positions
- Camera detects tag and calculates relative pose
- Robot position computed from known tag location
- Multiple tags improve accuracy through fusion
Benefits for FRC
- Corrects odometry drift automatically
- Enables accurate autonomous navigation
- Works regardless of starting position
- Provides absolute field coordinates
Multi-Camera Setup
Many top teams use multiple cameras: one facing forward for game pieces, and others positioned to always see AprilTags for continuous localization.
Recommended Approach for This Workshop
📷 Using Limelight
For this workshop, we'll use Limelight for its simplicity and reliability. The focus is on integrating vision data into your robot code, not configuring vision hardware.
What You'll Learn
- Reading vision data from NetworkTables
- Integrating AprilTag poses into odometry
- Using vision for target tracking
- Commanding turrets/shooters based on vision
Additional Resources
What's Next?
Up Next: Implementing Vision