Experimental

Sensor & Object Detection

Object Detection

In the object detection system, the team transitioned from using MobileNetv2 to YOLOv4. This change was made because MobileNetv2 was deemed less precise and flexible for handling all mission requirements. Therefore, YOLOv4 was implemented using the Darknet framework, customized to suit the Mandakini Zenith vessel. The comparison results between MobileNetv2 and YOLOv4 are as follows:

SensorObject
Hull Analysis Simulation

Sensor

For the sensor testing, the team evaluated the distance measurement accuracy of the LiDAR sensor using 10 distance variations with 20 cm intervals. The results showed a significantly improved maximum range of 12 m with LiDAR, compared to the HC-SR04 sensor used in Mandakini Hydra (2024), which had a maximum range of only 0.4 m.

Hull Analysis Simulation

Water Blast

The team tested the water blaster directly in the campus lake. The test was conducted by placing the nozzle at a distance of 1 meter from the target. The test results show that the water blaster can shoot as far as 2 meters at an angle of 45° from the water surface. The team also utilized a servo to move the nozzle towards the x and y axes.

Ball Launcher

Object delivery uses a flywheel-based system, where two wheels rotate in opposite directions to launch the ball. The test uses a 6 cm diameter wheel and a 1400 kV motor, with three PWM signalcontrolled motor speed variations to determine the maximum distance