Session: 18-01-02: AI Implementation in Industry - I

Paper Number: 149150

149150 - Development of Autonomous Cleaning Robot in Harsh Environment 

In the steel industry, raw material such as iron ore and coal that contains clumps, dust and powder is carried by belt conveyors and part of it falls to ground. Cleaning fallen raw material is currently done by human workers, but it is often physically demanding work in tight, low or dusty spaces around equipment. We developed an autonomous cleaning robot for harsh operating conditions such as high-temperatures and in dusty environments to reduce burden on human workers and improve safety and productivity. Our bulldozer-like cleaning robot autonomously follows cleaning paths and keeps its cleaning brush in contact with ground for collecting fallen clumps, dust and powder. It continuously calculates its position by self-localization estimation using a pre-built obstacle map and output data of LiDAR sensors. We overcame four issues for operating our robot steadily and continuously in harsh environments. First issue is a cooling system for protecting a robot’s internal electronic. When we want to use a robot in high temperature environment, we need to install a cooling system for protecting internal electronic from heat, but in dust environment, we can't use a conventional cooling fan system because it puts dust inside robot and damages internal electronics. We developed a unique cooling system which utilizes endotherm phenomena that occur when polymer material melts and water evaporates. We conducted test on top of coke oven and confirmed that the cooling system can prevent the temperature of internal electronics from exceeding upper limit of operating temperature limits. Second issue is noise data due to fuming or dust from LiDAR sensor output. LiDAR sensor emits laser light and measures distance by time it takes for reflected light returns. In dusty environment, part of emitted light is reflected by fuming or dust and this light generates noise data. We use multi-echo LiDAR sensors to reduce noise effects. Third issue is stuck avoidance for operational stability. We added error handling processes to a control flow of the robot. When the robot starts to travel cleaning path, it sets limit time for finishing traveling cleaning path. If it exceeds limit time, the robot judges traveling cleaning path as failure and takes actions for stuck avoidance. Forth issue is operational usability. We enabled operators to start cleaning on touch panel operation and check status on screen without typing commands on a computer. By virtue of overcoming these issues, our robot can work steadily and continuously for 4 hours with a simple operation. We named our robot, “GAZMASTAR^TM“ coined from the phrase “gather” and “master”. GAZMASTAR^TM is now in operation, performing heavy-duty cleaning tasks at our company.

Presenting Author: Yuki Takaki JFE steel corporation

Presenting Author Biography: Yuki TAKAKI majored in mechanical engineering at University of Tokyo and is working on development of autonomous traveling robots for labor saving in steel works as a senior researcher of JFE steel corporation. He likes to think about good use of robots while testing its on site.

Authors:

Yuki Takaki JFE steel corporation
Masaki Kobayashi JFE steel corporation
Koji Yamashita JFE steel corporation
Kyohei Ishida JFE steel corporation