Sensor Fusion, Mapping, Localization and Calibration of a Converted Autonomous Quad Bike

World is moving towards self-driving vehicles due to its’ safety and comfort. The primary motive of this work is to modify a manual All-Terrain quad bike to an autonomous one; by altering some mechanical features and sensor integration using Autoware.AI software. The LiDAR-Camera-IMU sensor system constitutes the primary sensors of the vehicle. The terrain mapping in this project is performed with Kalman Filter algorithm, known as Linear Quadratic Estimation (LQE). The bike is driven manually to create the required map data. Understanding the static and dynamic obstacles is done by this trial run of the vehicle. With Autoware.AI, the strict path and lanes can be defined that the vehicle will follow in the actual run. In the next stage, the actuators are programming to drive the bike based on the terrain conditions as analysed by the software system. SLAM (Simultaneous Localization and Mapping) algorithm is another handy tool for this project. It uses the Kalman Filter algorithm for its working, so the method is more accurate. At the same time, this property of SLAM makes it more difficult to operate; takes more computation power and time. Available resources put us in the limitation of using Kalman Filter only. After running the vehicle in the autonomous mode, it is important to check the deviation from the defined path and the accuracy of detecting the obstacles.