Performance Evaluation of Convolutional Networks on Heterogeneous Architectures for Applications in Autonomous Robotics
Fuente:
Redalyc
Saved in:
| Main Author: | |
|---|---|
| Format: | Artículo científico |
| Language: | en |
| Published: |
Instituto Tecnológico Metropolitano
2022
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1876446347295457280 |
|---|---|
| author | Joaquín Guajo |
| author_facet | Joaquín Guajo |
| contents | Performance Evaluation of Convolutional Networks on Heterogeneous Architectures for Applications in Autonomous Robotics Joaquín Guajo Cristian Alzate-Anzola Luis Castaño-Londoño David Márquez-Viloria Ingeniería on chip high system humanoid robot Humanoid robots find application in human-robot interaction tasks. However, despite their capabilities, their sequential computing system limits the execution of computationally expensive algorithms such as convolutional neural networks, which have demonstrated good performance in recognition tasks. As an alternative to sequential computing units, Field-Programmable Gate Arrays and Graphics Processing Units have a high degree of parallelism and low power consumption. This study aims to improve the visual perception of a humanoid robot called NAO using these embedded systems running a convolutional neural network. The methodology adopted here is based on image acquisition and transmission using simulation software: Webots and Choreographe. In each embedded system, an object recognition stage is performed using commercial convolutional neural network acceleration frameworks. Xilinx® Ultra96™, Intel® Cyclone® V-SoC and NVIDIA® Jetson™ TX2 cards were used, and Tinier-YOLO, AlexNet, Inception-V1 and Inception V3 transfer-learning networks were executed. Real-time metrics were obtained when Inception V1, Inception V3 transfer-learning and AlexNet were run on the Ultra96 and Jetson TX2 cards, with frame rates between 28 and 30 frames per second. The results demonstrated that the use of these embedded systems and convolutional neural networks can provide humanoid robots such as NAO with greater visual recognition in tasks that require high accuracy and autonomy. 2022 artículo científico 0123-7799 https://www.redalyc.org/articulo.oa?id=344270031007 https://www.redalyc.org/journal/3442/344270031007/ https://www.redalyc.org/journal/3442/344270031007/html/ https://www.redalyc.org/journal/3442/344270031007/344270031007.epub https://www.redalyc.org/journal/3442/344270031007/movil https://doi.org/10.22430/22565337.2170 en http://www.redalyc.org/revista.oa?id=3442 TecnoLógicas application/pdf Instituto Tecnológico Metropolitano TecnoLógicas (Colombia) Num.53 Vol.25 |
| format | Artículo científico |
| id | redalyc_344270031007 |
| institution | Redalyc |
| language | en |
| publishDate | 2022 |
| publisher | Instituto Tecnológico Metropolitano |
| spellingShingle | Performance Evaluation of Convolutional Networks on Heterogeneous Architectures for Applications in Autonomous Robotics Joaquín Guajo Ingeniería on chip high system humanoid robot Performance Evaluation of Convolutional Networks on Heterogeneous Architectures for Applications in Autonomous Robotics Joaquín Guajo Cristian Alzate-Anzola Luis Castaño-Londoño David Márquez-Viloria Ingeniería on chip high system humanoid robot Humanoid robots find application in human-robot interaction tasks. However, despite their capabilities, their sequential computing system limits the execution of computationally expensive algorithms such as convolutional neural networks, which have demonstrated good performance in recognition tasks. As an alternative to sequential computing units, Field-Programmable Gate Arrays and Graphics Processing Units have a high degree of parallelism and low power consumption. This study aims to improve the visual perception of a humanoid robot called NAO using these embedded systems running a convolutional neural network. The methodology adopted here is based on image acquisition and transmission using simulation software: Webots and Choreographe. In each embedded system, an object recognition stage is performed using commercial convolutional neural network acceleration frameworks. Xilinx® Ultra96™, Intel® Cyclone® V-SoC and NVIDIA® Jetson™ TX2 cards were used, and Tinier-YOLO, AlexNet, Inception-V1 and Inception V3 transfer-learning networks were executed. Real-time metrics were obtained when Inception V1, Inception V3 transfer-learning and AlexNet were run on the Ultra96 and Jetson TX2 cards, with frame rates between 28 and 30 frames per second. The results demonstrated that the use of these embedded systems and convolutional neural networks can provide humanoid robots such as NAO with greater visual recognition in tasks that require high accuracy and autonomy. 2022 artículo científico 0123-7799 https://www.redalyc.org/articulo.oa?id=344270031007 https://www.redalyc.org/journal/3442/344270031007/ https://www.redalyc.org/journal/3442/344270031007/html/ https://www.redalyc.org/journal/3442/344270031007/344270031007.epub https://www.redalyc.org/journal/3442/344270031007/movil https://doi.org/10.22430/22565337.2170 en http://www.redalyc.org/revista.oa?id=3442 TecnoLógicas application/pdf Instituto Tecnológico Metropolitano TecnoLógicas (Colombia) Num.53 Vol.25 |
| title | Performance Evaluation of Convolutional Networks on Heterogeneous Architectures for Applications in Autonomous Robotics |
| topic | Ingeniería on chip high system humanoid robot |
| url | https://www.redalyc.org/articulo.oa?id=344270031007 https://www.redalyc.org/journal/3442/344270031007/ https://www.redalyc.org/journal/3442/344270031007/html/ https://www.redalyc.org/journal/3442/344270031007/344270031007.epub https://www.redalyc.org/journal/3442/344270031007/movil https://doi.org/10.22430/22565337.2170 |