PaintPath: Defining Path Directionality in Maps for Autonomous Groud Vehicles

Directionality in path planning is essential forefficient autonomous navigation in a number of real-worldenvironments. In many map-based navigation scenarios, theviable path from a given pointAto pointBis not the sameas the viable path fromBtoA. We present a method thatautomatically incorporates preferred navigation directionalityinto a path planning costmap. This ‘preference’ is representedby coloured paths in the costmap. The colourisation is obtainedbased on an analysis of the driving trajectory generated bythe robot as it navigates through the environment. Hence,our method augments this driving trajectory by intelligentlycolouring it according to the orientation of the robot during therun. Creating an analogy between the vehicle orientation angleand the hue angle in the Hue-Saturation-Value colour space,the method uses the hue, saturation and value components toencode the direction, directionality and scalar cost, respectively,into a costmap image. We describe a costing function to be usedby the A* algorithm to incorporate this information to plandirection-aware vehicle paths. Our experiments with LiDAR-based localisation and autonomous driving in real environmentsillustrate the applicability of the method.