Abstract Obstacle avoidance is one of the main interests regarding path planning. In many situations (mostly those regarding applications in urban environments), the obstacles to be avoided are dynamical and unpredictable. This lack of certainty regarding the environment introduces the need to use local path planning techniques rather than global ones. A well-known method uses artificial potential fields introduced by Khatib. The Weyl potential definition have enabled to distinguish the dangerousness of an obstacle, however acceleration oscillations appear when the considered robot enters a danger zone close to an obstacle, thus leading to high energy consumption. In order to reduce these oscillations regarding this method, four alternative formulations for the repulsive field are proposed: corrective polynomials, tangential and radial components, Poisson potential and pseudo fractional potential. Their limitations will be explored and their performances will be compared by using criteria such as length and energy in a simulation scenario.