Energy efficiency is one of the most important parameters in transportation electrification. It allows to improve the production rate due to longer operation without charging or decrease the cost related to transportation. To provide collision-free operation in unknown or various environment, the local path planning algorithm should be considered. An Artificial Potential Field (APF) algorithm is commonly used for this task, however it provides unsmooth and oscillating motion of autonomous ground vehicle (AGV), and is prone to being trapped in a local minimum, e.g, dead-end. In such a case, the energy used to achieve the goal position is higher than necessary. In this paper, the energy-efficient local path planning algorithm is proposed. Future movement prediction has been introduced to APF to allow AGV to bypass obstacles in advance. A novel method for local minimum avoidance is introduced. It is based on placement of virtual obstacles called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">top quarks</i> in critical areas. These obstacles provide additional repulsive force for the APF based path planner. Considering the predicted stagnation-free path of the AGV, the new temporary goal for APF is selected. Such a combination allows to reduce traveled route length, improve its smoothness, and bypass local minima. The proposed Predictive Artificial Potential Field (PAPF) algorithm has been examined using Husarion ROSbot 2.0 PRO mobile robot, and the obtained results in form of videos are also attached as supplementary files. In comparison to original APF, the proposed path planning algorithm allows to reduce the used electric power by 21.4 %. PAPF provided a shorter path by up to 8.73 % and shorter time to reach the goal position by up to 40.23 %. The movement of the AGV is also much smoother in a case of usage of the proposed algorithm, and the proposed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">top quarks</i> -based local minimum avoidance mechanism allows to bypass the local minima.