Autonomous Navigation and Visual Analysis with Beam Robot Utkarsh Patel, Emre Hatay, Ghazal Zand May 10, 2017 Abstract We are improving the autonomous navigation of the Beam Robot, and also re- porting the detected human positions using the Kinect camera and the Beam robot. In this paper, we describe the problems we faced while transforming a closed plat- form telepresence robot into an autonomous robot, and also explain our approach on face detection. We successfully made the Beam robot navigate from the recre- ation center to the student center of Cleveland State University without any human involvement. For visual analysis, we developed a program using Cob People De- tection ROS package. Our future plan with this project is to have a team of fully autonomous Beam robots that can become permanent part of the Cleveland State University’s Engineering building. 1 Background We have used the Beam robot (Figure 1) for the entire project. Beam is a telepresence robot created by Suitable Technologies. Beam was originally developed by Willow Garage as Texai Remote Presence System, and spun out with most of Willow Garage team to build it as a commercial product. Our end goal with Beam is to completely transform it into a research robot for autonomous navigation and human robot interaction. Using Beam as a research platform has its own advantages and disadvantages. The advantage is that Beam has a human friendly hardware design, excellent driving control and cheaper price compared to other human-sized autonomous robots. Beam Robot has the same price as TurtleBot 2. The disadvantage is that it does not support ROS and it does not have a big user community. Beam is being used as a research robot by only few people in the world, and the company is not supporting any research projects. Beam runs on a Linux-based operating system with no source code provided. Beam does not come with autonomous functionality out of the box, but with the help of rosbeam package [3] and the open source nature of Linux, we were able to get full control of its motor. The rosbeam package makes it possible to send custom driving instruction to robot base, and to teleoperate Beam from any laptop. 2 Introduction Our project had two primary objectives. The first objective was to make Beam Robot navigate for longer distance without any human involvement. Before we started this 1