Autonomous Intersection Navigation Using Aerial Traffic Signal Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to navigate autonomous vehicles safely through intersections, particularly in determining the locations and statuses of traffic signals in real-time.
Innovation Solution
The system employs a combination of data from various sources, including oblique images, to identify traffic management features such as traffic lights and signs. This involves determining the latitude and longitude of intersection corners from aerial images, generating bounding boxes for traffic management features using machine learning, and calculating their coordinates through homography transforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional ground-based sensors are used to detect traffic signals, then the system can operate with simpler infrastructure, but the detection reliability deteriorates when signals are obscured by vegetation or other obstacles
Solution Approach 1:
The patent introduces aerial imagery (top-down perspective) as a new dimension for detecting traffic management features. By capturing images from above and processing them through homography transforms, the system can locate traffic signals that are obscured from ground-level views, thereby improving detection reliability without significantly complicating the overall system architecture
Solution Approach 2:
The patent uses aerial images as an intermediary data source to indirectly detect traffic management features. Instead of relying solely on direct line-of-sight from ground sensors, the system processes aerial imagery through coordinate transformation algorithms to infer the positions and states of traffic signals, acting as a mediator between unavailable direct observations and navigation decisions
2Measurement precision
If the autonomous vehicle uses multiple sensing modalities and complex processing to identify traffic signals, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent performs preliminary processing of aerial imagery to pre-identify and catalog traffic management features before the autonomous vehicle reaches the intersection. By using homography transforms and coordinate mappings in advance, the system creates a preprocessed map of traffic signal locations and states, reducing the real-time processing burden while maintaining high precision
Solution Approach 2:
The patent develops a universal coordinate transformation framework that can handle multiple types of traffic management features (traffic lights, signs, crosswalks) through a single homography transform process. This multi-functional approach allows the system to detect and process various traffic features using the same algorithmic foundation, improving precision without proportionally increasing system complexity
3Measurement precision
If the vehicle repositions itself to view traffic signals, then the detection accuracy improves, but the loss of time increases due to additional maneuvering
Solution Approach 1:
The patent uses aerial imagery to pre-identify and locate all traffic management features in the vicinity before the vehicle arrives at the intersection. By having this information available in advance through processed aerial data, the vehicle can directly navigate to optimal viewing positions without time-consuming trial-and-error repositioning, thus maintaining high detection accuracy while minimizing time loss
Data Source
AI summary
Systems and methods for navigating intersections autonomously or semi-autonomously can include, but are not limited to including, accessing data related to the geography and traffic management features of the intersection, executing autonomous actions to navigate the intersection, and coordinating with one or more processors and/or operators executing remote actions, if necessary. Traffic management features can be identified by using various types of images such as oblique images.