Sold
Available
IBL-26-1272

Self-Learning Based Control Method and Apparatus for Cleaning Robots

Listed on
2026-07-29
Robot Technology Wheeled/Tracked Robots Control/AI/SW
0.04
CI (SI)
★★★★★★★★★★
0.66
TR (N)
★★★★★★★★★★
0.06
MC
★★★★★★★★★★
Self-Learning Based Control Method for Cleaning Robots

This technology uses a camera to identify target substances on the floor and operates a learning mode that adjusts speed and suction intensity compared to the normal mode to determine the threshold speed and suction intensity required to remove specific substances. It is an algorithm that automatically switches to an optimized calibration mode when the same substance is detected, thereby controlling cleaning efficiency.

Existing cleaning robots operate at a constant speed and suction intensity without considering the environment or the characteristics of specific substances, leading to residual debris and reduced energy efficiency.

This technology applies a control logic that extracts threshold values by sequentially performing a normal mode and a learning mode (variable speed/intensity) during the Nth cleaning cycle, and dynamically switches to a calibration mode suitable for the target substance identified by the camera during the N+1th cleaning cycle. Applicable to industrial robots and automation systems, it improves cleaning efficiency and optimizes operation by learning the optimal movement speed and suction intensity for different target materials through unsupervised learning.

Key Features:
  • A step of operating a camera-equipped cleaning robot under normal mode conditions with a reference speed or reference suction intensity for a first set time during an Nth cleaning process.
  • A step of applying learning mode conditions to the cleaning robot, which gradually vary the speed or increase the intensity compared to the normal mode conditions, and operating it for a second set time.
  • A step of extracting the threshold speed or threshold suction intensity at which a target substance that was not removed during the first set time begins to be removed, and acquiring a calibration mode containing the extracted information.
  • A self-learning based control method for a cleaning robot, comprising a step of switching to and operating in the calibration mode acquired during the Nth cleaning process when a target substance of the same type as the previously identified target substance is detected by the camera.
DGIST
Jin-Hyo Yoon | Jeong-Ho Yang
Document
Date of application:
2014-05-02
|
Patent registration number:
10-1526117
Industry
robot•automation
Technology
Robotics
Optics•Sensor
Country
Korea
Family Patent

N/A

Price
가격협의
Subscribe to our newsletter to receive the latest patent information faster than anyone else.
← Back to list