Our next-generation vehicle concept features a trimaran hull form optimized for low resistance and high stability. The slender center hull minimizes wave-making resistance, while the side hulls (amas) create a wide buoyancy lever for superior initial stability.

Carbon Fiber Construction: Selected for its high rigidity-to-weight ratio and corrosion resistance. The modular internal volume includes watertight compartments for batteries and electronics, ensuring easy service access and precise center-of-gravity management.

• Trimaran Form: Low resistance, high stability.
• Structure: Lightweight carbon fiber composite.
• Specs: ~15 cm Draft, ~35 kg Displacement.

The mechanical propulsion layout is designed to combine rapid cruise capability with precise lateral maneuvering. The thruster configuration is physically integrated into the hull to maximize control authority.

Thruster Configuration:

Lateral (Blue Robotics T200): Mounted at the bow and stern for high-precision docking and rotation. Capable of delivering up to 6.7 kgf of thrust at peak load.

Main Drive (Blue Robotics T500): The primary forward propulsion units. Designed for speed, generating up to 16.1 kgf of thrust per unit, enabling high-speed transit and dynamic positioning in strong currents.

• Hybrid Layout: Combines T200 (Precision) and T500 (Power).
• Performance: High thrust-to-weight ratio for agile response.

Our ROS 2 (Humble) software stack runs on the NVIDIA Jetson Orin Nano Super and is primarily developed in Python. The system continuously acquires sensor data, processes and logs information in parallel, and executes mission-level decision making. Based on real-time and recorded inputs, the autonomy layer performs path planning and triggers actions according to mission requirements.

• ROS 2 Humble–based modular autonomy architecture
• Parallel sensor processing, logging, and mission execution
• Path planning and mission logic running on Jetson Orin Nano

The ZED 2i stereo camera provides synchronized RGB and depth information for buoy and obstacle detection. Depth maps and semantic detections are converted into local cost representations that support safe navigation in structured waterways and competition environments.

• Real-time RGB + depth from ZED 2i stereo camera
• Buoy, obstacle, and marker detection using computer vision pipelines
• Depth-aware risk mapping for navigation and mission logic

The control and monitoring interface runs on the NVIDIA Jetson onboard computer and is accessible through a standard web browser. Developed using HTML, CSS, and JavaScript, it provides real-time visualization of telemetry, mission status, and perception outputs via WebSocket-based communication.

• Runs on Jetson Orin Nano, accessible via web browser
• Real-time telemetry and mission data via WebSocket
• Integrated live camera feed and map-based route tracking
• Extensible architecture for future control and sensor visualizations

This simulation framework is designed to test and validate stereo camera–based perception and object detection algorithms developed for the unmanned surface vehicle (USV) within the Gazebo simulation environment.

A virtual stereo camera integrated into the simulated vehicle provides synchronized image streams, which are acquired through the ROS 2–Gazebo interface and processed using a Python-based computer vision pipeline.

The developed object detection models are executed on simulated image data to identify obstacles and mission-related objects. Depth and distance information extracted from stereo imagery are utilized to estimate the relative positions of detected objects in the environment.

This simulation infrastructure enables safe and repeatable testing of algorithms under different environmental and mission scenarios before deployment on real hardware, allowing performance evaluation and observation of failure cases without physical risk.

• Synchronized image streaming via a virtual stereo camera in Gazebo
• Real-time image data transfer through ROS 2 integration
• Integration of Python-based object detection algorithms
• Depth and distance estimation from stereo imagery
• Pre-deployment validation of perception algorithms and software components
• Repeatable testing under diverse mission scenarios and environmental conditions