Threshold-type obstacle recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robots face challenges with low success rate when navigating threshold-type obstacles, particularly threshold-type obstacles, such as door thresholds, table and chair legs, and other protruding objects, due to inaccurate identification and orientation recognition.
Innovation Solution
A robot control method that integrates target axis data such as angular velocity, angular acceleration, and body pitch to identify threshold-type obstacles, and uses point cloud data clustering to determine the orientation, enabling accurate navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional obstacle recognition methods are used, then the robot can detect obstacles, but the success rate of navigating threshold-type obstacles remains low
Solution Approach 1:
The patent segments the obstacle detection task into multiple components: point cloud data collection from multiple sensors, clustering analysis to identify threshold-type obstacles, skeleton line extraction to determine orientation, and separate processing for different obstacle types. This segmentation allows each component to be optimized independently, improving both detection reliability and measurement precision.
Solution Approach 2:
The patent transitions from traditional 2D sensor data to 3D point cloud data, adding spatial dimensionality to obstacle recognition. By using point cloud clustering in three-dimensional space and extracting skeleton lines with orientation information, the system achieves more accurate identification and orientation recognition of threshold-type obstacles.
2Measurement precision
If multiple sensor data types are integrated for obstacle identification, then the accuracy of identification improves, but the device complexity increases
Solution Approach 1:
The patent employs a universal point cloud processing framework that can handle data from multiple sensor types (laser radar, depth camera, etc.) through a common clustering and analysis pipeline. This multi-functional approach allows the same processing algorithms to work with different sensor inputs, improving identification accuracy without proportionally increasing system complexity.
Solution Approach 2:
The patent introduces point cloud data as an intermediary representation that bridges different sensor types and the obstacle recognition algorithm. By converting various sensor inputs into a unified point cloud format, the system simplifies data integration and reduces the complexity of handling heterogeneous sensor data directly.
3Adaptability or versatility
If the robot uses traditional obstacle avoidance methods, then it can navigate general obstacles, but it fails to accurately recognize and pass through threshold-type obstacles
Solution Approach 1:
The patent implements dynamic obstacle recognition that adapts to different obstacle types in real-time. By using point cloud clustering and skeleton line extraction, the system dynamically identifies threshold-type obstacles and calculates their orientation, then adjusts the robot's navigation strategy accordingly. This dynamic adaptation improves both versatility and orientation recognition accuracy.
Solution Approach 2:
The patent changes the recognition parameters from simple obstacle presence detection to detailed point cloud analysis with clustering and skeleton line extraction. By transforming the detection parameters to include spatial distribution, density, and orientation information, the system achieves accurate recognition of threshold-type obstacles while maintaining adaptability to different obstacle categories.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The present application relates to a robot control method, comprising: triggering a pass-through threshold-type obstacle command based on a recognition result of a threshold-type obstacle; obtaining an orientation of the threshold-type obstacle in response to the pass-through threshold-type obstacle command; obtaining a direction of travel of a robot relative to the orientation of the threshold-type obstacle; and controlling the robot to pass through the threshold-type obstacle based on the direction of travel of the robot relative to the orientation of the threshold-type obstacle.