In modern day robotic applications, the use of cloud computing is being considered as a viable option giving rise to development of cloud robotics environment. Robots are also being developed to operate under an organized framework of robot operating system (ROS) for flexibility and better integration with robots of different types. In this work, an investigation is presented on the development of a mobile robotic platform and its integration in a ROS enabled cloud robotics environment. A mobile robotic platform was built with different sensors including a depth camera, integrated with Arduino and raspberry pi for interfacing the sensors, the drive system and on-board local processing of signals and with wireless communication capability for transmission and receiving data.
The robot was operated using ROS framework within a cloud robotics network. Two issues of robot operation in ROS enabled cloud environment, namely, latency and data integrity were investigated for the developed robot under different operating conditions. The system was tested for baseline connectivity and under low bandwidth environment and performance was found to be satisfactory in the areas of latency and data integrity. Ongoing and future extensions are proposed to integrate this current robot with other existing robots within the ROS enabled cloud robotics environment.
Raspberry Pi (shown in Fig. 7) is a smaller than a phone computer that can be utilized as a desktop PC. It is little in any case, sufficiently capable to do consistent word processing, spread sheet investigation and other general exercises. It can likewise do nonstandard task like being a media server or go about as a smart TV. The Raspberry Pi was developed in the United Kingdom by the Raspberry Pi Foundation with the intention of promoting the teaching of basic computer science in schools and developing countries. The original Raspberry Pi and Raspberry Pi 2 are manufactured in several board configurations through licensed manufacturing agreements with Newark element14 (Premier Farnell), RS Components and Egoman. The hardware is the same across all manufacturers.
Cloud robotics and cloud computing in general, offers the concept of utility computing; instead of buying computational resources to own, these resources can be provided for use on the short-term, as-needed basis (Armbrust, et al. 2010). With the use of cloud robotics, robots are able to store, interact, and solve more complex information in a data center. This way there is a decreased dependence needed in the middle ware of a system.
Physically, the devices were connected to a Dell Power Connect 6248 Ethernet switch via category 6 twisted pair cables. The Power Connect switch was segmented into three virtual local access networks (VLANs) to create the three aforementioned networks (shown in Figure 12) and prevent traffic from crossing over the networks. For wireless connectivity on the Testbed network, a Linksys E2700 wireless router was used to provide a Wi-Fi access point for remote systems.
The most important point of view in this phase is to find a position in which the sensor operates at its best, and manages to produce reasonable details of what has been framed. The usage of the GPUs of graphics cards implies some compatibility limits, and in fact along with the software a compatibility list is supplied with the various graphics cards.
EXPERIMENTAL RESULTS AND DISCUSSIONS
Testing the bandwidth of the network was necessary to determine the limits of the network. From the observation it was clear that the wired network performed exceptionally well compared to the wireless network, but the wireless network was a better fit.
One clear conclusion is that these tests are better done in the night time with a short time scale. The best results came from the 1:06 time scale which wasn’t too long and the results were great.
CONCLUSIONS AND RECOMMENDATIONS
A mobile robotic system, Arlobot, worked well without significant issues of network latency and data integrity. The robot was integrated with a ROS enabled cloud robotic environment. The Arlobot had local processors of Arduino Mega and Raspberry Pi. The Robot Operating System was used as the main source of interaction for the robot to-cloud in the cloud network. The Raspberry Pi provided the interface between the cloud and the robot in the network. The network performance and data integrity were investigated when the Arlobot communicated with the cloud network. Baseline performance was established with the robot in the network. Experiments were conducted to test the network performance under different conditions.
Recommendations for Future Work
Due to the time constraints, the investigations were limited to the low bandwidth data. In the next phase, investigations with high-bandwidth data would be considered. The Arlobot had a lot of room for improvement although everything was running correctly, another part that should have been viewed is possibly using another Arduino for the simple fact that the Arduino board kept burning out.
Source: Georgia Southern University
Author: Theodore C. Smith