InfiniteInferno

Land on a project idea and validate whether our planned hardware stack can realistically support it.

What we did

• Brainstormed multiple hardware project directions for the 9-week build. ARGUS — an autonomous drone that:

  • Visually tracks a person/object (bounding-box based detection)

  • Pursues the target

  • Maps hand gestures to onboard actions (e.g., fire suppression or target-specific actions)

  • Tracking, pursuit, and gesture-mapped actions treated as modular features

  • Allows partial implementation instead of a rigid full-stack dependency

  • Pixhawk + Arduino UNO Q + camera stack

  • Considered compute requirements, power draw, and payload constraints

• Compared Arduino UNO Q vs Raspberry Pi across:

  • Compute capability
  • Power consumption
  • Camera/ISP support
  • Real-time control ability
  • Software ecosystem maturity

Problems and blockers

• Neither Arduino UNO Q nor Raspberry Pi (standalone) can run heavy real-time detection models (e.g., full YOLO) at high FPS.

  • Lightweight models (YOLOv8n / MobileNet-SSD)

  • Reduced resolution for real-time performance on a moving drone

  • Very new board with limited documentation

  • Sparse examples for MAVLink integration and computer vision workloads

  • Lacks a real-time MCU

  • Time-critical actions (gesture triggers, actuation) require an additional microcontroller

Decisions

• Core: Person/object tracking + autonomous pursuit via Pixhawk

• Stretch goals: Gesture-mapped actions and fire/target-response mechanisms

• Deferred final choice of companion computer (UNO Q vs Raspberry Pi) to Week 1 after hands-on testing

  • All autonomous flight testing must be tethered or caged initially

Next week

• Pixhawk + companion computer + camera

• Real-time detection performance

• Latency and control loop feasibility

• Integration gaps between vision and flight control

Links

• Code:
• Photos / CAD: