Design of Automated Construction System for Modular Structures based on Parameterized Learning Automata

Abstract

This work presents the development and imple- mentation of an autonomous construction system which uses a holonomic mobile robot for constructing three-dimensional modular structures from parts of different size. The main contribution of this paper is to propose a novel architectural framework for planning the construction task by reducing the total traveled distance and learning time with respect to the state of the art. A high-level planner based on Parameterized Learning Automata (PLA) and $\mathrm {A}^{\ast }$ algorithms uses environmental feedback to generate the robot actions for accomplishing the task. The robotic platform is equipped with a magnetic gripper and also a RGB-D camera and other additional sensors to aid the control and navigation of the robot. The simultaneous mapping and robot location are performed using the Real-Time Appearance-Based Mapping (RTAB-Map) algorithm. The RTAB-Map algorithm uses three-dimensional image information and odometry data to create the map environment and also to estimate the pose of the robot. Simulation and experimental trials have been performed to assess the effectiveness of the proposed robotic assembly system.