Obstacle Detection Using Semantic-Constrained Depth Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intelligent driving systems face poor obstacle recognition accuracy due to sensor blind spots and insufficient generalization capabilities, leading to inaccurate detection of general obstacles like stones and fallen trees, which can result in collisions.
Innovation Solution
An obstacle detection method that constructs a depth map using structure constraint information based on semantic types such as ground, wall, and sky information, followed by processing to identify traveling and non-traveling regions, enhancing depth map accuracy and obstacle recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sensor-based obstacle detection is used, then the system can detect common obstacles, but it fails to accurately recognize general obstacles like stones and fallen trees due to sensor blind spots and insufficient generalization capability
Solution Approach 1:
The patent transitions from 2D image data to 3D spatial understanding by constructing depth maps. This dimensional transformation enables the system to perceive obstacle height and spatial relationships, allowing accurate detection of general obstacles like stones and fallen trees that have varying heights and positions, thereby resolving the limitation of traditional 2D-based detection methods.
Solution Approach 2:
The patent changes the parameter representation from simple 2D coordinates to 3D depth information. By introducing depth maps that encode spatial distance and height parameters, the system gains the ability to distinguish obstacles based on their vertical and horizontal positions, improving both measurement precision and adaptability to various obstacle types.
2Measurement precision
If depth map construction is performed without structure constraint information, then the processing is simpler, but the depth map accuracy is insufficient leading to poor obstacle recognition
Solution Approach 1:
The patent incorporates structure constraint information as feedback during depth map construction. By using semantic segmentation results and spatial relationship constraints to guide and refine the depth estimation process, the system iteratively improves depth map accuracy while maintaining a manageable construction complexity through algorithmic optimization.
Solution Approach 2:
The patent introduces structure constraint information as an intermediary element between raw image data and final depth maps. This intermediary layer provides spatial and semantic guidance that bridges the gap between simple image processing and accurate 3D reconstruction, improving depth map accuracy without requiring overly complex direct processing methods.
Data Source
AI summary
The present disclosure relates to obstacle detection methods, apparatuses, systems, and computing devices. One example method includes obtaining an image, then constructing a depth map of the image based on structure constraint information, and after completing depth map construction, processing the depth map to obtain a region identifier map including a plurality of regions, where each of the plurality of regions is a traveling region or a non-traveling region. The structure constraint information includes a semantic type of each sample in the image, and the non-traveling region is considered as an obstacle.