The codes for implementation of tasks have been written in C++. Microsoft Visual Studio has been used as the software framework for VARUN. The software is designed to run in decentralised multi-threaded agent architecture, with the threads handling pressure sensor, acoustics, cameras, control system, IMU each performing input and output in a continuous loops. GUIs have been developed to allow users to configure each of vehicle’s sensor systems, as also to adjust control parameters. The onboard computer is used to interface the different sensors, as also to run the codes.

Image Processing

The concepts of image analysis and processing have been implemented using Intel’s Open CV (Open Source computer vision library) for real time performance. The vertical poles of the gate are determined by ‘Sobel’ edge detection technique, with Hough Transform to approximate thelines over the detected edges. To detect the flashing red light, the RGB image taken be forward looking camera is converted to HSV (Hue, Saturation and Value) color space. A binary image is found by applying proper thresholds after examining hue and brightness values, and the vehicle is ‘moved’ to the centroid in the centre of the screen. Pipeline detection is done be the camera angled at 45 degrees. The HSV format again is used to examine hue values. Hough lines are drawn and their orientation gives the direction of travel. The downward looking cameras are used to detect the bins. Canny edge detector with two thresholds is applied to determine strong edges and weaker edges connected to strong edges. Consequentially an enclosed bin is obtained and the centroid of the resultant blob determines the required centre for the marker. 

Control Programming

The primary aim of the control system is to achieve control over 3 degrees of freedom since the stable mechanical design overcomes roll and pitch. It follows simple motion control algorithms using PID feedback implemented in C++.  The target output is shifted from program parameters to vector inputs form vision and sound codes, based on task trigger basis. User interfaces developed specifically for the purpose provide to adjust the PID parameters.

Wireless Access System

A remote control for monitoring and manual control of VARUN is a significant addition to the team’s work-out plan for the second generation vehicle. This is achieved by a mix of Ethernet and wireless 802.11g networks. A trailing router is used to communicate with VARUN and an off-board computer, even when the vehicle is submerged in water.