A Feasible Architecture Of Real-Time Collision Avoidance And Path Planning For Semi-Autonomous Unmanned Ground Vehicle (Ugv)

An Unmanned Ground Vehicle (UGV) is a vehicle that is on contact with ground and operates without human on-board. UGVs are widely used for mission-based applications that are often hazardous or inconvenient for humankind. Practically, UGVs are equipped with multiple devices like sensors and on-board camera to enable self-navigation and fulfil the mission requisite. In this project, the UGV combines two main elements, which are collision avoidance system (CAS) and path planning system (PPS). Costing is highly concerned for this project, therefore a low-cost ultrasonic sensor (HC�SR04) is used as a distance sensor for obstacle detection and avoidance purpose. Meanwhile, an Arduino Mega 2560 and ArduPilot Mega 2.6 (APM) are used as microcontrollers for obstacle avoidance and path optimization purpose respectively. Vehicle navigation is based on point-to-point basis and the waypoints are set in Mission Planner software that works alongside APM. In this project, the vehicle is programmed to maneuver at lower speed between 2 to 4 km/h. The maneuver is assisted with ultrasonic sensors (front and rear) sense and avoid the obstacles within 3 meters range. The ultrasonic sensors are programmed in such a way that it sweeps the sensors that the field of view of 180° angle both front and rear of the vehicle. Path planning is the key element for unmanned vehicle. Therefore, APM is used for navigation before and after approaching obstacles. Path planning algorithm in this project provides the best path to avoid the obstacle and create temporary new waypoints within split seconds. This practice makes the UGV more authentic and proximately imitable with human sensors and its responses