Sensor-Fused Path Planning Without High-Precision Vector Maps

The patent segments the road detection task into two parts: using simple image processing to detect road line positions and using distance sensor data to measure actual distances. This segmentation allows the system to achieve high precision without requiring complex vector maps, as each component performs a simplified function that can be accurately executed independently.

The patent creates a virtual road model by copying and integrating data from simple image processing results and distance sensor measurements. This virtual model replicates the functionality of high-precision vector maps but is generated dynamically from inexpensive sensors, eliminating the need for complex pre-existing map data while achieving comparable or superior accuracy.

The system performs self-service by using the vehicle's own distance sensors and cameras to generate accurate path information in real-time. Instead of relying on externally created and maintained vector maps that require continuous manual updates and government cooperation, the system autonomously creates its own navigation data from onboard sensors, eliminating ongoing manufacturing and maintenance costs.

The patent changes the fundamental parameters of the system from using pre-existing high-precision map data to using real-time sensor measurements. By shifting from static map data to dynamic sensor-based measurement, the system achieves high precision without the manufacturing burden of creating and maintaining detailed vector maps, as the measurement process itself becomes the source of accuracy.

The patent merges image processing results with distance sensor measurements to achieve accurate path detection. The image processing provides road line positions while the distance sensor provides accurate range data. By combining these two simple components, the system achieves both low device complexity and high measurement precision, as the strengths of each component compensate for the weaknesses of the other.

The patent introduces a data fusion algorithm as an intermediary that processes and integrates information from image processing and distance sensing. This intermediary component coordinates the data from both sources, reconciling their different measurement types and formats into unified, accurate path information, thereby enabling the simple components to work together effectively.