Artificial potential fields revisited: advances, limitations, and hybrid solutions in modern robotics

Abstract Local path planning algorithms for autonomous mobile robots (AMRs) have gained significant attention in recent years. One of the most commonly used approaches is the artificial potential field (APF) algorithm. Its popularity is based on its simplicity, low computational cost, and efficiency. Nevertheless, the classical APF has some well-known drawbacks. First, the APF generates oscillations near obstacles. Moreover, it is sensitive to parameter tuning. The APF algorithm also exhibits poor performance in narrow passages and stagnates in local minima. Therefore, its applicability in modern AMR systems is limited. This paper presents a review of recent applications and modifications of the APF algorithm. First, the original APF algorithm is presented. Next, recent modifications of repulsive strategies, additional potentials, and adaptation strategies are discussed. The integration of APF with additional mechanisms, such as prediction, optimization, is also presented. Moreover, this review includes shared-control and teleoperation schemes in which APF is used as a collision-avoidance layer. The paper concludes with a discussion of open challenges and promising research directions. This review aims to provide a structured reference for practitioners and a roadmap for future APF-based research.