Path Planning using Model Predictive Controller based on Potential Field for Autonomous Vehicles Zahra Elmi Autonomous Systems Laboratory Department of Computer Engineering Hacettepe University Ankara, Turkey E-mail: Zahra.elmi@gmail.com Mehmet Önder EFE Autonomous Systems Laboratory Department of Computer Engineering Hacettepe University Ankara, Turkey E-mail: onderefe@hacettepe.edu.tr Abstract— In recent decades, one of the challenging problems is path planning for autonomous vehicle in dynamic environments with along static or moving obstacles. The main aim of these researches is to reduce congestion, accidents and improve safety. We propose an optimal path planning using model predictive controller (MPC) which automatically decides about the mode of maneuvers such as lane keeping and lane changing. For ensuring safety, we have additionally used two different potential field functions for road boundary and obstacles where the road potential field keeps the vehicle for going out of the road boundary and the obstacle potential field keep the vehicle away from obstacles. We have tested the proposed path planning on the different scenarios. The obtained results represent that the proposed method is effective and makes reasonable decision for different maneuvers by observing road regulations while it ensures the safety of autonomous vehicle. Keywords—Path Planning, Autonomous Vehicles, Model Predictive Control, Sequential Quadratic Programming, Artificial Potential Field. I. INTRODUCTION Nowadays, the autonomous vehicle has been becoming an important research in vehicle engineering. In autonomous vehicle or intelligent unmanned vehicle, the safety of vehicle is a critical factor. The development of autonomous vehicle technology has a significant effect on reducing traffic, accidents and relieving the stress of motorists. The automobile manufacturers such as Tesla Motors and several internet companies have decided to develop fully autonomous vehicles in order to avoid collides in the roads, but these vehicles are very complex systems and include sensing systems [1], path planning systems, trajectory tracking systems and suspension systems [2], [3]. Although, there are many researches on the path planning and tracking to avoid collision for unmanned vehicle and other robots, it is not easy to utilize these approaches directly to avoid collision scenarios. Because, these vehicles are only able to move in its stable limitations and handling capability in a constrained environment. Therefore, to solve collision avoidance problem in the road, it is necessary to consider the other moving vehicles that have their own motion properties. Early researches on the path planning for autonomous vehicle date back to the 1980s that focus on the computation of optimal time and collision-free path from a given point to another. Many approaches are presented for path planning that can be categorized in two classifications: conventional and heuristic approaches. Conventional approaches are referred to common methods such as sampling based algorithms and grid- based methods. The sampling based algorithms such as rapidly exploring random trees algorithm and its variants, perform by sampling of configuration to connect the initial configuration with the target [4]. Grid-based algorithms have mapped free space to set of cells and then try to solve a problem of graph search. The major shortcoming of the classic approaches is that they do not consider the dynamics of the vehicle. In contrast, the controller approaches such as Model Predictive Control (MPC) consider the model of system to predict the evaluation of system states and has been widely used in a variety of systems [5]. Due to its capabilities to overcome physical constraints, MPC has been used to path planning and tracking for the autonomous vehicle. MPC based approaches are a problem of optimization and solve it for finding an optimal open-loop control sequence which minimizes the objective function by satisfying all the state and input constraints. Many conventional approaches are proposed for path planning methods such as A* heuristic search, Visibility graph, Voronoi diagram and Potential field [6]. The motion of the autonomous vehicle and the sum of used forces are combined by the potential field method [7]. This method is consisting of two forces, one is the attractive force to move the vehicle towards goal or destination and the other is repulsive force to avoid collisions with obstacles. However, an inherent problem of traditional potential field methods is the formation of local minimum that prevents from arriving the vehicle to target. In this paper, we developed a new potential field method based on boundary road and the vehicle’s kinematic model to generate a collision free path for autonomous vehicle. The proposed algorithm can be updated to obtain a collision free path along with static and dynamic obstacles in real time. The reminder of paper is organized as follows. In section II,