Autonomous Navigation Using Image Segmentation and Sparse Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating due to the vast volume of data from sensors, cameras, and maps, which can limit navigation accuracy and efficiency, and traditional mapping technologies require significant storage and updates.
Innovation Solution
A navigation system using trained networks to process images from cameras, identifying traffic lights and contextual information to determine navigational actions, integrating with sparse maps for efficient navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional mapping technology is used for navigation, then comprehensive map coverage is achieved, but data storage requirements and update complexity increase significantly
Solution Approach 1:
The patent segments the navigation system into multiple specialized neural networks: a first trained network for traffic light state detection and a second trained network for contextual information processing. This segmentation allows each network to focus on specific aspects of navigation, reducing the overall data processing burden and improving efficiency without sacrificing navigation accuracy.
Solution Approach 2:
The patent extracts only the essential features needed for navigation decisions from the full captured image. Instead of processing entire high-resolution images, the system identifies and processes specific segments containing traffic lights and contextual information, significantly reducing data volume while maintaining navigation reliability.
2Measurement precision
If vast volumes of sensor data are collected and analyzed, then navigation decision accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The system performs preliminary processing by pre-training specialized neural networks offline to recognize traffic light states and contextual information patterns. During real-time operation, these pre-trained networks quickly process incoming data without requiring extensive computational resources, thus maintaining high decision accuracy while minimizing processing time.
Solution Approach 2:
The patent replaces traditional mechanical image processing methods with neural network-based processing. The trained networks automatically extract relevant features and make decisions based on learned patterns, eliminating the need for complex algorithmic processing and significantly reducing computation time while maintaining or improving accuracy.
3Reliability
If multiple neural networks are used for specialized processing, then navigation accuracy is improved, but system complexity increases
Solution Approach 1:
The patent designs the trained networks to perform multiple functions within their specialized domains. The first trained network handles both traffic light detection and state classification, while the second network processes various contextual information types. This multi-functionality reduces the need for additional separate components, managing system complexity while maintaining high navigation accuracy.
Data Source
AI summary
A navigation system for a host vehicle may include at least one processor comprising circuitry and a memory. The memory may include instructions that when executed by the circuitry cause the at least one processor to receive an image representative of an environment of the host vehicle, to identify a first segment of the image, to provide the first segment to a first trained network, the first trained network being configured to generate a first output indicative of a state of a traffic light, to identify a second segment of the image, to provide the second segment to a second trained network, the second trained network being configured to generate a second output indicative of a proposed navigational action, to determine, based on both the first output and the second output a planned navigational action and to cause the host vehicle to take the planned navigational action.