Mobile Robot Image Selection for Accurate Object Attribute Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional mobile robots lack accurate object attribute recognition, leading to unreliable object avoidance and inefficient cleaning performance, as they struggle to differentiate between objects they can climb and those that pose a hazard, such as electric wires or human hair.
Innovation Solution
A mobile robot equipped with an image acquisition unit for continuous image capture, a sensor unit for object detection, and a controller that selects images based on moving direction and speed to recognize object attributes using machine learning, allowing for precise object recognition and adjustment of travel patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional sensors (infrared or ultrasonic) are used for object detection, then the mobile robot can sense presence and distance of objects, but the robot cannot accurately determine object attributes to differentiate between climbable objects and hazardous objects
Solution Approach 1:
The patent combines multiple sensor types (infrared sensor, ultrasonic sensor, and camera) into an integrated sensing system. The infrared and ultrasonic sensors detect object presence and distance, while the camera captures images for attribute recognition. This merging of sensors allows the robot to achieve accurate object attribute recognition without requiring a completely new sensing system.
Solution Approach 2:
The patent introduces a camera as an intermediary device that bridges the gap between simple object detection and complex attribute recognition. The camera captures images that serve as intermediate data, which are then processed by machine learning algorithms to determine object attributes. This intermediary approach enables accurate recognition without directly complicating the primary sensing system.
2Reliability
If the mobile robot uses only height-based judgment for object avoidance, then the control logic is simple, but the robot cannot differentiate between safe objects (like low thresholds) and hazardous objects (like electric wires or human hair)
Solution Approach 1:
The patent performs preliminary image capture and attribute analysis before the robot makes avoidance decisions. The camera continuously captures images of detected objects, and machine learning algorithms pre-analyze these images to determine object attributes. This preliminary action allows the robot to reliably distinguish between climbable objects and hazardous objects before executing avoidance maneuvers.
Solution Approach 2:
The patent replaces simple mechanical height-based judgment with an intelligent image recognition system using machine learning. Instead of relying solely on physical measurements and predetermined height thresholds, the system uses computer vision and AI algorithms to automatically recognize object attributes, significantly improving reliability while managing complexity through software-based solutions.
3Measurement precision
If the mobile robot captures images at object sensing time, then the system responds quickly to detected objects, but the selected image may not contain the object due to the robot's movement
Solution Approach 1:
The patent performs preliminary image capture before the robot moves to the object's position. The camera continuously captures images in advance, and the system selects appropriate pre-captured images that contain the detected objects. This preliminary action ensures that images are captured when the robot is still in position, guaranteeing object inclusion while maintaining quick response times.
Solution Approach 2:
The patent dynamically adjusts the timing of image capture based on the robot's movement state and object detection results. The system flexibly selects images captured at different time points, choosing the most appropriate pre-captured image that contains the object. This dynamic approach balances the need for quick response with the requirement for accurate object capture.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances object recognition accuracy, enabling reliable object avoidance and improved cleaning efficiency by accurately determining object attributes and adjusting movement patterns accordingly.
Implementation Method 1
the ultrasonic sensor emits an ultrasound wave having a predetermined period and, in response to ultrasonic waves reflected by an object, determines a distance to the object based on a difference between the time to emit the ultrasonic waves and the time for the ultrasonic waves to return to the object
Implementation Method 2
The infrared sensor determines presence of an object and a distance to the object based on an amount of light reflected by the object
Data Source
Figure 1~3
Figure 4~6
Figure 7~9
AI summary
A mobile robot in one general aspect of the present invention includes: a travel drive unit configured to move a main body; an image acquisition unit configured to acquire a plurality of images by continuously photographing surroundings of the main body; a storage configured to store the plurality of continuous images acquired by the image acquisition unit; a sensor unit having one or more sensors configured to sense an object during the movement of the main body; and a controller is configured to, in response to the sensing of the object by the sensor unit, select an image acquired at a specific point in time earlier than an object sensing time of the sensor unit from among the plurality of continuous images based on a moving direction and a moving speed of the main body, and recognize an attribute of the object included in the selected image acquired at the specific point in time, and accordingly, it is possible to acquire image data enabling high accuracy in object attribute recognition, and accurately recognize an attribute of an object.