Year-end Sale % OFF 
Abstract
This paper presents the implementation of a mobile robotic arm simulation whose task is to order different objects randomly distributed in a workspace. To develop this task, it is used a Faster R-CNN which is going to identify and locate the disordered elements, reaching 99% accuracy in validation tests and 100% in real-time tests, i.e. the robot was able to collect and locate all the objects to be ordered, taking into account that the virtual environment is controlled and the size of the input image obtained from the workspace to be entered to the network should be 700x525 px.
Highlights
The applications of robotics in different areas have had an exponential growth in recent years, such as in medicine with surgical assistants [1,2], in the agricultural sector [3], in the industry for high-risk tasks [4] or even in tasks of daily life with assistance robots [5,6] or service in the cleaning area [7]
In [8] an adaptive model based on neural networks is used for the planning of trajectories and evasion of obstacles, or in other cases, Deep Learning techniques are used, such as the Convolutional Neural Networks (CNN) [9], where through these it is possible to perform the remote control of a mobile robot by means of commands made by a user with different hand gestures, in such a way as that executes certain action [10]
CNNs have been combined with localization techniques, so that, apart from recognizing the object, it locates it through a Region of Interest (RoI), creating a so-called Region-Based CNN [13], which consists of using region proposal algorithms in conjunction with CNN, they have a long execution time, making their use in real-time applications inefficient
Summary
The applications of robotics in different areas have had an exponential growth in recent years, such as in medicine with surgical assistants [1,2], in the agricultural sector [3], in the industry for high-risk tasks [4] or even in tasks of daily life with assistance robots [5,6] or service in the cleaning area [7]. In [8] an adaptive model based on neural networks is used for the planning of trajectories and evasion of obstacles, or in other cases, Deep Learning techniques are used, such as the Convolutional Neural Networks (CNN) [9], where through these it is possible to perform the remote control of a mobile robot by means of commands made by a user with different hand gestures, in such a way as that executes certain action [10]. CNNs have been combined with localization techniques, so that, apart from recognizing the object, it locates it through a Region of Interest (RoI), creating a so-called Region-Based CNN (or R-CNN) [13], which consists of using region proposal algorithms in conjunction with CNN, they have a long execution time, making their use in real-time applications inefficient. In recent years the CNN architectures have been improved, through the implementation of new combinations to reduce processing times, as shown in [14], where Region Proposal Networks are used, obtaining a network called Faster R-CNN, which is faster, and improves location accuracy
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Online and Biomedical Engineering (iJOE)
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
