In recent years, the autonomous vehicle domain has grown enormously, and a lot of research is currently being conducted in this field. The autonomous vehicles available in the market are Autonomous Underwater Vehicles (AUVs), Unmanned Aerial Vehicles (UAVs) and Autonomous Ground Vehicles (AGVs). The primary requirement for an autonomous vehicle is a path planning system. The path planning system provides the preplanned path to carry out the designated tasks and guidance in the field operation to tackle uncertainties faced by the vehicle.

The primary application of autonomous vehicles is inspection. The path planning system requirement for inspection applications is unique because the vehicle needs to cover the whole survey region, and this unique path plan is called the Coverage Path Plan (CPP). The CPP is required to have the least trajectory length and turns to save energy and time spent for the operation, and an optimal CPP for an autonomous vehicle is developed in this work to satisfy this requirement. The polygon extracted from the workspace boundaries is provided as input to the developed CPP and is generally concave type. The developed path planner decomposes the input polygon into many convex polygons and then unifies this decomposed polygon to reduce the polygon count. For this decomposition and unification operation, a formulation based on polygon's sweep line orientation is developed and solving this formulation by Dijkstra Algorithm (DA) ensures the least number of turns in the developed algorithm's output path. The order in which the vehicle visits the polygons (obtained from decomposition and unification operation) affects the overall trajectory length. So, the Ant Colony Optimisation (ACO) technique that optimises the order of polygons to be visited by the vehicle is developed, ensuring that the overall trajectory length is optimised.

Even though CPP provides the preplanned path for the operation, obstacles are faced by the vehicle while in operation. For evading these obstacles, a collision avoidance algorithm is also developed that uses a tree (viable movements the vehicle can make from the current location) generated on the workspace to evaluate an obstacle-free path (branches of the tree) in the presence of static obstacles. The tree's growth is constrained by the vehicle's kinematic constraints, such as maximum turning angle and linear velocity. The efficient tree generation method proposed in this work generates branches only in the direction of vehicle's viable movement. The collision avoidance algorithm uses a heuristics-based A* algorithm which solves the tree for an obstacle-free path based on the cost value defined at the tree nodes. The developed algorithms in this work are versatile and can be used in many autonomous vehicles that operate in a two-dimensional plane.