An Intelligent Space for Mobile Robot Localization Using a Multi-Camera System

Abstract

This paper describes an intelligent space, whose objective is to localize and control robots or robotic wheelchairs to help people. Such an intelligent space has 11 cameras distributed in two laboratories and a corridor. The cameras are fixed in the environment, and image capturing is done synchronously. The system was programmed as a client/server with TCP/IP connections, and a communication protocol was defined. The client coordinates the activities inside the intelligent space, and the servers provide the information needed for that. Once the cameras are used for localization, they have to be properly calibrated. Therefore, a calibration method for a multi-camera network is also proposed in this paper. A robot is used to move a calibration pattern throughout the field of view of the cameras. Then, the captured images and the robot odometry are used for calibration. As a result, the proposed algorithm provides a solution for multi-camera calibration and robot localization at the same time. The intelligent space and the calibration method were evaluated under different scenarios using computer simulations and real experiments. The results demonstrate the proper functioning of the intelligent space and validate the multi-camera calibration method, which also improves robot localization.

Particle filter robot localisation through robust fusion of laser, WiFi, compass, and a network of external cameras

“…We are interested in exploring simple functions such as averaging or weighted averaging, and complex functions that may capture relationships among sensors that arise from training data. Although our camera observation model performed well for our purposes, we also want to explore the use of models that can provide information about the exact pose of the robot [36,60].…”

“…When fused with other sensors, even this simple model can provide sufficiently accurate and robust pose estimates. It is expectable that camera models able to provide more accurate pose estimates [35][36][37][38]60] improve these results even more.…”

Survey on artificial intelligence based techniques for emerging robotic communication

“…Zaouche et al [59] Zhong et al [41] Skobelev et al [50] Rampinelli et al [48] Shibata [42] Elleuch et al [21] Schleich et al [30] Ghouti et al [38,39]…”

A flexible human detection service suitable for Intelligent Spaces based on a multi-camera network

“…The system is also able to control actuators like a robot. The cameras are used as the main sensors of the environment, similarly to the project described in Rampinelli et al 43 In order to acquire high-level understanding of the environment, we designed a software infrastructure to tackle the task of processing and analyzing data extracted from the distributed cameras in real-time.…”