Details
Original language | English |
---|---|
Title of host publication | Embedded Computer Systems |
Subtitle of host publication | Architectures, Modeling, and Simulation - 24th International Conference, SAMOS 2024, Proceedings |
Editors | Luigi Carro, Francesco Regazzoni, Christian Pilato |
Publisher | Springer Science and Business Media Deutschland GmbH |
Pages | 138-154 |
Number of pages | 17 |
ISBN (electronic) | 978-3-031-78377-7 |
ISBN (print) | 9783031783760 |
Publication status | Published - 28 Jan 2025 |
Event | 24th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2024 - Samos, Greece Duration: 29 Jun 2024 → 4 Jul 2024 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 15226 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (electronic) | 1611-3349 |
Abstract
Autonomous driving systems require performant and reliable perception, though they only possess limited computational resources, which places a high priority on the efficiency of the underlying algorithms. Radar sensors play an important role in this context, because they provide data in the form of sparse point clouds, which can be stored and processed in a condensed and efficient manner. However, this sparsity is often overlooked in the design of perception algorithms, such as convolutional object detection networks. In this work we investigate how sparse submanifold convolutions can be used to exploit this sparsity to drastically reduce the computational complexity of a CNN-based radar object detector. To this end, we propose an efficient implementation of submanifold convolutions on a vertical vector processor architecture called V2PRO, which is emulated on an FPGA board. Benchmarks on the public nuScenes dataset and an internal dataset show, that the sparse models provide competitive detection performance, while achieving average speedups of up to 27x over their dense counterparts on the considered vector processor. Finally, the sparse model deployed on the FPGA is integrated into a measurement vehicle with three front-facing high-resolution radars, to demonstrate real-time online radar object detection running at 15 Hz.
Keywords
- FPGA, Neural network accelerators, Perception, Radar object detection, Sparse CNN, Vector Processor
ASJC Scopus subject areas
- Mathematics(all)
- Theoretical Computer Science
- Computer Science(all)
- General Computer Science
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Embedded Computer Systems: Architectures, Modeling, and Simulation - 24th International Conference, SAMOS 2024, Proceedings. ed. / Luigi Carro; Francesco Regazzoni; Christian Pilato. Springer Science and Business Media Deutschland GmbH, 2025. p. 138-154 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 15226 LNCS).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Radar Object Detection on a Vector Processor Using Sparse Convolutional Neural Networks
AU - Köhler, Daniel
AU - Meinl, Frank
AU - Blume, Holger
N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
PY - 2025/1/28
Y1 - 2025/1/28
N2 - Autonomous driving systems require performant and reliable perception, though they only possess limited computational resources, which places a high priority on the efficiency of the underlying algorithms. Radar sensors play an important role in this context, because they provide data in the form of sparse point clouds, which can be stored and processed in a condensed and efficient manner. However, this sparsity is often overlooked in the design of perception algorithms, such as convolutional object detection networks. In this work we investigate how sparse submanifold convolutions can be used to exploit this sparsity to drastically reduce the computational complexity of a CNN-based radar object detector. To this end, we propose an efficient implementation of submanifold convolutions on a vertical vector processor architecture called V2PRO, which is emulated on an FPGA board. Benchmarks on the public nuScenes dataset and an internal dataset show, that the sparse models provide competitive detection performance, while achieving average speedups of up to 27x over their dense counterparts on the considered vector processor. Finally, the sparse model deployed on the FPGA is integrated into a measurement vehicle with three front-facing high-resolution radars, to demonstrate real-time online radar object detection running at 15 Hz.
AB - Autonomous driving systems require performant and reliable perception, though they only possess limited computational resources, which places a high priority on the efficiency of the underlying algorithms. Radar sensors play an important role in this context, because they provide data in the form of sparse point clouds, which can be stored and processed in a condensed and efficient manner. However, this sparsity is often overlooked in the design of perception algorithms, such as convolutional object detection networks. In this work we investigate how sparse submanifold convolutions can be used to exploit this sparsity to drastically reduce the computational complexity of a CNN-based radar object detector. To this end, we propose an efficient implementation of submanifold convolutions on a vertical vector processor architecture called V2PRO, which is emulated on an FPGA board. Benchmarks on the public nuScenes dataset and an internal dataset show, that the sparse models provide competitive detection performance, while achieving average speedups of up to 27x over their dense counterparts on the considered vector processor. Finally, the sparse model deployed on the FPGA is integrated into a measurement vehicle with three front-facing high-resolution radars, to demonstrate real-time online radar object detection running at 15 Hz.
KW - FPGA
KW - Neural network accelerators
KW - Perception
KW - Radar object detection
KW - Sparse CNN
KW - Vector Processor
UR - http://www.scopus.com/inward/record.url?scp=85218469713&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-78377-7_10
DO - 10.1007/978-3-031-78377-7_10
M3 - Conference contribution
AN - SCOPUS:85218469713
SN - 9783031783760
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 138
EP - 154
BT - Embedded Computer Systems
A2 - Carro, Luigi
A2 - Regazzoni, Francesco
A2 - Pilato, Christian
PB - Springer Science and Business Media Deutschland GmbH
T2 - 24th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2024
Y2 - 29 June 2024 through 4 July 2024
ER -