StrathVoyager Student Team Technical Report to NJORD Challenge 2023

Louvros, Panagiotis and Zhao, Wang and Lee, Paul and Arcidiacono, Mauro Francisco and Gorska, Kaja and Tsoumpris, Charalampos and Ohol, Nilesh and Wright, Marvin Stuart and Troll, Moritz and Iqbal, Muhammad and Ait Ameur, Mohamed Adlan and Stefanou, Evangelos and Jimoh, Isah Abdulrasheed and Aung, Myo Zin and Dai, David and Jia, Laibing and Theotokatos, Gerasimos (2023) StrathVoyager Student Team Technical Report to NJORD Challenge 2023. University of Strathclyde.

[thumbnail of Njord_Competition_2023_Technical_Report]
Preview
 Text. Filename: Njord_Competition_2023_Technical_Report.pdf
Final Published Version
License: Strathprints license 1.0
Download (2MB)| Preview

Abstract

Starting from basic design and operational concepts on paper, as modular ASV was designed and built within approximately 12 months on a budget of around £5000. The boat participating in the Njord competition features a catamaran design comprised of two custom designed NPL hulls. The catamaran platform was designed to be fully modular and is constructed with 3D-printed hulls and carbon fibre decks. The platform is propelled using a static dual-thruster configuration. Computation is facilitated by an Nvidia Jetson, a Raspberry Pi 4B, and an Arduino MEGA. The system is powered by two custom-made 5S BMS 21V Li-Ion battery packs. The navigation sensor system consists of an Ouster OS32 3D Lidar with a built-in IMU, a Stereovision Zed Mini depth camera, an Adafruit Ultimate GPS Module PA1616D, and an Adafruit BNO055 Absolute Orientation Sensor. Further internal sensors, including voltage, current sensors, as well as a temperature sensor, are integrated. The ASV connects wirelessly to a remote control via an nRF2401 module, and to a laptop for operation monitoring via Holybro Telemetry Sik radio. Autonomous operation is enabled through sensor fusion of camera and Lidar information to identify environmental features, such as waypoints, and to generate virtual waypoints for the autonomous control system. The autonomous control system consists of an in-house developed deep reinforcement learning (DRL) algorithm that enables Line of Sight operation, as well as obstacle avoidance. As a backup, a general PID LOS waypoint tracking controller is also implemented in parallel. Extensive lab and pond testing has been carried out to develop and refine the system. Key features and innovations of the StrathVoyager ASV include: • Fully modular 3D printed and carbon fibre ASV. • Depth camera and 3D lidar sensor fusion for autonomous control. • DRL-based autonomous control.

CORE (COnnecting REpositories)