An Efficient LIDAR Sensing System for Self Driving Cars

The Self-driving car is an autonomous robot that navigates to its destination without human operator. The aim of this project is to make an efficient lidar sensing system for Self-driving cars that is capable of mapping its surroundings, navigating through the path, and reaches the destination automatically. This is done by implementing Simultaneous Localization and Mapping (SLAM) algorithm on a collected series of lidar scans using pose graph optimization to build the map of the environment. The robot recognizes a previously visited place through scan matching and establishes one or more loop closures along its moving path. The SLAM algorithm utilizes the loop closure information to update the map and adjust the estimated robot trajectory. To plan a path through an environment effectively a probabilistic roadmap (PRM) algorithm uses a network of connected nodes to find an obstacle-free path from a start to an end location. Nodes are connected based on the obstacle locations specified in the Map. The results are demonstrated on a low-cost Autonomous RC Robot that uses ROS (Robotic Operating System) software library, version of ROS Kinetic booted on Raspberry Pi interfaced with YDLidar X2 in the front top portion of it. This low-cost autonomous bot emerges with features like SLAM (Simultaneous Localization and Mapping), Path planning, and Path Following which can reach the destination automatically after its destination is fixed on the Map.