Comparative Performance Analysis of Navigation Algorithm and Deep Learning Application: Different Infrastructure and Cloud Robotics




Bhaskaran, Divya

Journal Title

Journal ISSN

Volume Title



Robotic applications have been evolving from time to time from a remote control, autonomous, real-time to self-learning. Advancement in coding or software platforms is not the only reason for the development but for hardware infrastructure also. Thus for any robot to perform with utmost efficiency, the choice of correct software and hardware is vital. A heavy robotic algorithm on a low-level hardware or onboard computation may produce poor results.

The main problem in many robotic applications is latency due to poor hardware selection even if the software is optimized. These scenarios especially occur in vision-based and deep learning applications which demands heavy computation making real-time applications difficult to implement. A GPU system or a cloud infrastructure can best suit these algorithms and overcome the latency issue and increase the performance efficiency.

The idea is to select two complex robotic applications and implement in different infrastructure and compare the results and illustrate that the applications implemented on a cloud or GPU hardware performs better than the onboard computing. Two popular robotic applications, one with localization and mapping - RTAB map (VSLAM) and the other with Natural Language Processing - Sentiment Analysis of Movie Reviews (deep learning application) are chosen for this experiment. Both the applications are implemented in the different scenario such as on board, GPU and Cloud infrastructure and tested for best results.


This item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.


Cloud Robotics, Machine Learning, VSLAM



Electrical and Computer Engineering