Method, apparatus and storage medium for detecting small obstacles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current obstacle detection systems for indoor mobile robots, such as cleaning robots, face difficulties in recognizing small obstacles and depressions due to challenges in mounting cameras at low positions and maintaining a horizontal state, which affects the robot's operation and can lead to entanglement or smearing issues.
Innovation Solution
A method involving the acquisition of a 3D point cloud from image data, extraction of a ground region, calculation of a ground projection point cloud, and morphological analysis to identify small obstacles within a set height range, utilizing inertial measurement unit data and learning models for accurate obstacle detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the camera is mounted at a low position to detect small obstacles, then the obstacle detection capability is improved, but it becomes difficult to maintain a horizontal state
Solution Approach 1:
The patent introduces an intermediary computational process (point cloud processing, ground plane detection, height difference calculation) between the camera and the obstacle detection task. This mediator processes the image data to compensate for the non-horizontal camera state, enabling accurate small obstacle detection without requiring the camera to be physically mounted horizontally.
Solution Approach 2:
The patent transforms the detection parameters by calculating height differences relative to the ground plane rather than using absolute pixel coordinates. This parameter transformation allows the system to detect small obstacles even when the camera is not in a horizontal state, as the relative height measurements remain valid regardless of camera orientation.
2Measurement precision
If the camera is mounted at a low position, then small obstacle detection is improved, but the robot may experience entanglement or smearing issues
Solution Approach 1:
The patent performs preliminary detection of small obstacles and depressions before the robot encounters them. By identifying these hazards in advance through image processing and height difference analysis, the robot can plan its path to avoid entanglement or smearing issues, thereby maintaining operational reliability while still using a low-position camera.
Solution Approach 2:
The patent converts the potential harm of operating close to the ground (risk of entanglement and smearing) into a benefit by using the low-position camera to detect small obstacles and depressions that higher cameras would miss. This early warning capability allows the robot to avoid these hazards through intelligent navigation, turning the operational risk into enhanced detection capability.
3Device complexity
If traditional obstacle detection methods are used, then the system complexity is low, but small obstacles and depressions cannot be effectively recognized
Solution Approach 1:
The patent introduces height difference as an additional dimension for obstacle detection. By calculating the vertical distance between detected objects and the ground plane, the system can identify small obstacles and depressions that traditional 2D image processing would miss, significantly improving recognition capability without requiring complex multi-sensor systems.
Solution Approach 2:
The patent segments the image processing task into distinct stages: ground plane detection, object detection, height difference calculation, and classification. This segmentation allows the system to effectively recognize small obstacles and depressions through a series of manageable processing steps rather than requiring a single complex detection algorithm.
Data Source
AI summary
A method for detecting small obstacles includes: acquiring a first 3D point cloud corresponding to image data acquired by a cleaning robot; extracting, from the first 3D point cloud, a second 3D point cloud of a ground region; extracting, from the second 3D point cloud, a third 3D point cloud having a height value in a set height range; calculating a ground projection point cloud of the third 3D point cloud; and, determining morphologically-connected regions in the ground projection point cloud, and using a morphologically-connected region having an area less than a preset value as a region where a small obstacle is located. By effectively recognizing the ground, acquiring the ground projection point cloud and processing the point cloud by image processing, accuracy and timeliness of the algorithm can be improved compared with directly processing discrete 3D point clouds.