Autonomous vehicles are a trending research topic which involves a number of different problems. This research work considers the problem of path planning, while taking into account a specific and important traffic scenario – determination of the path in order to perform obstacle avoidance during cornering. In this research work, a novel hybrid path planning approach which consists from two classical planning approaches – the Bezier curves and the Rapidly-exploring random trees, is proposed and experimentally tested. The entire development process of the hybrid path planning approach is described. The experimental testing of the proposed approach was performed while using an actual prototype of the autonomous vehicle. During the experimental testing, different positions of the obstacle, different radius of the corner arc and different angles of the corner arc were considered. Approximately 200 different scenarios were considered and experimentally tested. The results of the experimental testing showed that the general performance of the proposed hybrid path planner is reliable – the planned paths are safe, continuous, and technically feasible.