A Novel Algorithm for a Low-Cost, Curvature-Continuous Smooth Path with Multiple Constraints on a Cost-Assigned Flat Map

Mobile robots are extensively utilized across various fields, with path planning consistently representing a core and pivotal area of research. Path planning is essential for enabling the efficient navigation of robots within complex environments. In reality, the terrain on which the robot operates is non-uniform, resulting in varying costs associated with different areas due to differing terrains and materials. Practical tasks often necessitate traversing a series of landmark points to fulfill specific requirements. Furthermore, considerations related to control and dynamics frequently require setting minimum line segment lengths between curves and maximum curve curvatures to ensure the successful execution of the path. The objective of this paper is to find a low-cost path with continuous curvature on a map with an assigned cost, which passes through all the given landmark points while avoiding obstacles, and satisfies the minimum length of the line segments between the curves and the maximum curvature constraints of the curves. We propose an innovative path planning method that solves the limitations of traditional algorithms by considering map cost, curvature continuity, and other factors by establishing a collaborative mechanism between global coarse search and local fine-tuning. The method is divided into two stages: In the first stage, the graph structure is constructed by generating points on the map, and uses Dijkstra’s Algorithm to obtain the connection order of the landmark points. In the second stage, which builds on the previous stage and processes landmark points sequentially, the key points of the path are generated using our proposed Smooth Beacon Reconnection (SBR) algorithm. A low-cost path meeting the requirements is then obtained through fine-tuning. The smooth path generated by this method is verified on multiple maps and demonstrates superior performance compared to traditional methods.