This is a static obstacle avoidance algorithm that each of us conceived and developed as a Path Planning team member for MACARON 5.0. My idea is to convert the D algorithm to local path generation and use it.*
Unlike other navigation algorithms that sequentially navigate the path from its current location to the target point, d star algorithm calculates spatial information for a given entire grid map, starting from the target point. Select the optimal path from the target point to the starting point based on the cost values for the entire grid map thus created.
I envisioned an algorithm that uses d*algorithm to generate local routes and avoid static obstacles. Given information about lanes and center lines, create possible global routes from your current location to your target point. During the drive, an obstacle appears on the route, reflecting this, creating a new route.
Basically, as an algorithm that calculates information about the entire grid map and generates a global path based on it, we were able to benefit from the computational speed.
By giving different cost values for the center line and the lane, we envisioned an algorithm to try to change the lane without invading the center line if the current lane does not have a driving route.
In addition, I conducted hyper parameter tuning to stop if there is an obstacle on the current route and if lane change is impossible.
The longer the path, the longer the algorithm will take. And in the competition, I have to restart the algorithm if the car gets out of driving. The use of D star algorithm in such competitions was unsuitable and unfortunately was not used.