Vehicle Track Generation via Risk Map Binarization and Thinning

The patent segments the continuous risk map into discrete grid cells with specific risk values. By dividing the navigation space into a grid structure where each cell contains risk information, the system transforms a continuous optimization problem into a discrete pathfinding problem on a graph, significantly reducing computational complexity while maintaining track accuracy.

The patent performs preliminary action by pre-calculating and storing risk values in each grid cell before track generation. The risk map is prepared in advance with evaluation items computed for all cells, allowing the track retrieval process to simply follow the pre-computed risk gradients rather than performing exhaustive calculations during actual navigation.

The patent substitutes the mechanical computation process with an information retrieval process. Instead of performing complex numerical optimization calculations to find the minimum sum of evaluation items, the system uses pre-computed risk maps and simple pathfollowing algorithms that retrieve the optimal track by following risk gradients, dramatically reducing processing time while maintaining accuracy.

The risk map is computed in advance before actual navigation occurs. All evaluation items and risk values are calculated beforehand and stored in the map structure, so that during real-time operation, the system only needs to retrieve and follow the pre-computed information rather than performing time-consuming calculations.