Driving pathfinding of unmanned autonomous ground vehicle using measurement data diffusion

The state of road changes quite often due to the automobiles and pedestrians when the main driving unit controls the unmanned autonomous vehicle along the planned path. The vehicle acknowledges of whether there are obstacles on the driving path using a sensor array and creates the new driving path and adaptively updates route to control the vehicle. This research proposes a novel way to find the possible driving path by diffusing the measurement data collected by the sensor array which contains the unscaled info of the detected obstacles. With the possible driving field, we can recognize whether the current path would be affected by the obstacles, and also possibly create a new driving path to avoid them. Using the driving map created by this way, we made a new driving path applying A* algorithm and tested on a unmanned autonomous vehicle (i.e., converted KIA Soul). As a result, after creating the new driving path, we were able to carry out the avoidance driving safely at low speed, also the vehicle drives swiftly and smoothly when we modified the avoidance path within the possible driving field.