Intersection Contention Area Detection Without HD Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving systems face challenges in accurately detecting and classifying intersection contention areas, especially in complex urban environments, due to reliance on detailed annotations for deep neural networks and the failure to handle occlusions and transient conditions without up-to-date HD maps.
Innovation Solution
A deep neural network-based system processes sensor data to detect and classify intersection contention areas in real-time, using signed distance functions and post-processing techniques to determine 3D world-space locations, allowing vehicles to navigate intersections safely without prior knowledge or detailed maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional systems use multiple separate deep neural networks to detect and combine various features (traffic lights, stop signs, vehicle positions, lanes, etc.) for intersection detection, then the detection accuracy may be improved, but the device complexity and computational intensity increase significantly
Solution Approach 1:
The patent combines multiple separate deep neural networks that detect different features (traffic lights, stop signs, vehicle positions, lanes, free-space boundaries) into a single unified neural network. This unified network receives sensor data and directly outputs intersection areas and classifications, eliminating the need to separately detect and combine multiple features. The merging reduces device complexity and computational overhead while maintaining detection accuracy.
2Measurement precision
If detailed annotations are required to train deep neural networks for complex intersections, then the measurement precision of intersection detection improves, but the ease of manufacture and scalability decrease
Solution Approach 1:
The patent changes the output parameters of the neural network from detecting individual features to directly predicting intersection areas and their classifications. By modifying the network architecture and training objectives to output semantic segmentation masks with intersection classifications (e.g., pedestrian crossing, vehicle entering, vehicle exiting, no lane), the system reduces the need for detailed manual annotations of multiple features while maintaining classification accuracy.
3Reliability
If map-based solutions are used to analyze intersections by comparing detected features to pre-stored HD maps, then the reliability of intersection understanding improves, but the adaptability to transient conditions and unavailable maps decreases
Solution Approach 1:
The patent enables the vehicle's perception system to independently detect and classify intersections using sensor data and a unified neural network, without relying on external HD maps. The system processes sensor data to generate intersection areas and classifications directly, making the vehicle self-sufficient in intersection understanding. This approach maintains reliability while improving adaptability to transient conditions and areas where maps are unavailable or outdated.
Data Source
AI summary
In various examples, live perception from sensors of a vehicle may be leveraged to detect and classify intersection contention areas in an environment of a vehicle in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute outputs—such as signed distance functions—that may correspond to locations of boundaries delineating intersection contention areas. The signed distance functions may be decoded and/or post-processed to determine instance segmentation masks representing locations and classifications of intersection areas or regions. The locations of the intersections areas or regions may be generated in image-space and converted to world-space coordinates to aid an autonomous or semi-autonomous vehicle in navigating intersections according to rules of the road, traffic priority considerations, and/or the like.