Robots, obstacle detection methods, and related devices
By combining ultrasonic sensors and depth cameras and fusing RGB-D information, an obstacle prediction probability map is created, which solves the shortcomings of RGB-D sensors in detecting transparent objects, enables effective detection of objects such as glass, and improves the robot's obstacle avoidance capabilities.
CN116310773BActive Publication Date: 2026-07-07SHENZHEN PUDU TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN PUDU TECH CO LTD
- Filing Date
- 2021-12-08
- Publication Date
- 2026-07-07
AI Technical Summary
Technical Problem
Existing RGB-D sensors perform poorly in detecting transparent and specular objects, making it difficult for robots to avoid obstacles effectively, especially failing to detect glass objects.
Method used
By combining ultrasonic sensors with depth cameras, and fusing RGB-D and ultrasonic information, a grid area obstacle prediction probability map is created. Historical ultrasonic data is used for anomaly detection and data correction, and point cloud image processing is combined to achieve effective detection of transparent objects.
Benefits of technology
It enables effective detection of transparent objects, improves the accuracy and reliability of robot obstacle avoidance, and enhances the ability to recognize objects such as glass.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116310773B_ABST
Abstract
This application relates to the field of obstacle detection technology. Embodiments of this application provide a robot, obstacle detection method, and apparatus. The method involves acquiring ultrasonic data from a target area using an ultrasonic sensor; acquiring a depth image of the target area using a depth camera; determining a first data set based on the ultrasonic data; determining a second data set based on the depth image; creating a target map in a first coordinate system, the target map containing the target area and dividing the target map into a preset number of grid areas; determining the predicted probability of an obstacle existing in each grid area based on the first and second data sets; and designating grid areas in the target map whose predicted probability of an obstacle existing is greater than or equal to a given threshold as a first area, which represents the location of the obstacle on the target map. Embodiments of this application can detect transparent or semi-transparent objects, achieving effective obstacle detection by fusing RGB-D and ultrasonic information.
Need to check novelty before this filing date? Find Prior Art