Abstract
Although there are many kinds of robots with various purposes, hard and dangerous work is generally among their tasks. In this vein, we have been studying robots for stairs cleaning. Floors are basically two-dimensional, but stairs are three-dimensional. Action for a stairs cleaning robot is clearly more complex, and it must avoid dangerous situations like falling. In addition, there are many kinds of staircase designs. Stair conditions also vary as to location (indoor/outdoor) and status of use. Thus, for stairs cleaning, the robot is desired to work with an advanced learning function. In this post, we explain a 6-DOF robot that is designed and built to clean stairs. We propose control regulation based on reinforcement learning.
