The third and the final iteration of the ASV was designed with an emphasis placed on drag reduction. The availability of better manufacturing techniques and a higher budget allocation to the mechanical domain enabled the hull to be designed with emphasis on streamlining. The hull was made with fiberglass by hand-laying over a mold crafted using foam board. The data gathered previously using the second prototype enabled efficient exploitation of the trimaran configuration. Thruster and cable penetrator mounts were included in the hull to ensure rigidity and ease of fixing. Other fixtures include handles for ease of transportation and harness points for deployment. The superstructure of Trishul includes three sensors – GPS, camera and Li-Dar. These three sensors are mounted on a single sensor pod to free up space for other payload on the deck.
Electronics in Trishul were fabricated keeping in mind a modular design. Each module is fixed on its mount with the ability to be quickly detached for troubleshooting or replacement. The purpose of every mount is to provide stability, a rigid support as well as power to the corresponding boards. All PCBs were fabricated and optimized for minimum size, heat dissipation and error-free functioning giving the power distribution system an output efficiency of 98.5%.
Trishul uses a sophisticated navigation algorithm involving localization through a Garmin 19x HVS GPS module along with the implementation of a Kalman Filter on the IMU data received using a Sparton AHRS-8 module to achieve Dead Reckoning and further increase localization accuracy. The Communication System allows access to critical system commands in a server over Wi-Fi accessing JSON messages using HTTP requests. The Software suite also comprises of a Mission Planner and Tracking system to allow Trishul to always be aware of what task is being executed and update its status. The Computer Vision system in the boat is designed to accomplish the preliminary and mission tasks and to assist in navigation of Trishul. The algorithms were developed and tested using the MATLAB and Simulink platform due to their rich library and availability of vision critical functions. After a number of testing in real time conditions the thresholds for buoys and shape detection have been optimized and the control signals were hence mapped. Efforts have been done to eliminate the errors due to factors such as ambient light, reflection, etc. and to optimize the execution speed of the System.