This space is dedicated to our Hackathon Project: SHARK USV Autonomous Navigation & Intelligent Obstacle Avoidance

<aside> 👋 Welcome to the team! Our mission is to design and protoype an autonomous navigation system that…

• Consumes and interprets real-time LiDAR scanning data as well as a video feed to detect, classify, and track nearby vessels, buoys, and static hazards
• Implements swarming algorithms to produce smooth, natural waypoint plans to a given destination and updates them dynamically as new sensor data arrives
• Predicts obstacle trajectories using temporal tracking and proactively adjusts course to avoid conflicts using emergent navigation behaviors
• Maintains robust navigation during challenging maritime conditions including sensor noise, water reflections, and partial occlusions

📢 Communication

• Email contact: george@deepdock.xyz
• Coordination/Preparation: Konstantinos
• Hackers: Deno, Konstantinos +2
• Mentor: George </aside>

Technical Approach

Perception Layer:

• 360° scanning LiDAR system (TF03 rangefinder on servo actuator) → to be designed and built up during hackathon
• GoPro Hero7 camera with object detection neural network (YOLO/MobileNet/custom model)
• Real-time obstacle detection and clustering
• Sensor fusion combining visual semantics with geometric ranging data
• Temporal tracking for moving object velocity estimation
• Sensor fusion with GPS/IMU for robust localization

Navigation Layer:

• Bio-inspired boids algorithm for multi-objective path planning
• Weighted behavioral rules (obstacle avoidance, goal seeking, path smoothing)
• Dynamic speed control based on obstacle proximity
• Emergency stop logic for critical collision scenarios
• Class-aware safety margins: larger buffers for boats/kayakers, smaller for buoys/markers
• Dynamic speed control based on obstacle proximity and threat level
• Emergency stop logic for critical collision scenarios

Control Layer:

• MAVLink interface to ArduRover autopilot
• Real-time heading and velocity commands
• Failsafe integration with manual override capability

Success Criteria

Minimum Viable Demonstration:

• SHARK autonomously navigates to a waypoint 50+ meters away
• Camera detects and classifies at least 2 object types (e.g., "boat", "buoy")
• LiDAR provides ranging to detected objects
• System fuses vision + LiDAR for enhanced obstacle awareness
• Maintains safe clearance margins (>5 meters from hazards)
• Generates smooth, boat-appropriate motion paths

Stretch Goals:

• Detect and avoid moving obstacle with trajectory prediction
• Demonstrate class-specific safety behaviors (wider berth for boats vs. buoys)
• Complete multi-waypoint survey mission through obstacle field
• Achieve >90% object classification accuracy on maritime objects
• Real-time visualization showing fused LiDAR + vision perception
• Predict obstacle trajectories and proactively re-route
• Demonstrate recovery from GPS degradation using dead reckoning

Provided Resources

Hardware (All sensors and equipment will be provided):

• SHARK USV platform (1.7m hull with propulsion and steering ) → already working on manual
• Orange Cube autopilot with ArduRover firmware
• TF03 LiDAR rangefinder with 360° servo scanning system
• Here4 GPS module for global positioning
• Herelink remote control system with video telemetry
• GoPro Hero7 black camera for mission documentation
• LiPo battery system with full day endurance
• Waterproof electronics enclosures
• NVIDIA Jetson Nano/Orin Super for advanced processing
• Raspberry Pi 3+ SENSE for basic processing

Additional hardware can be purchased on Saturday 11.10.2025 if needed.