Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimization of a Line Detection Algorithm for Autonomous Vehicles on a RISC-V with Accelerator

María José Belda, Katzalin Olcoz, Fernando Castro, Francisco Tirado

TL;DR

The paper addresses real-time line-detection for autonomous vehicles under onboard power and latency constraints. It proposes a heterogeneous architecture that pairs a RISC-V core (Rocket or BOOM) with the Gemmini matrix-multiplication accelerator, and systematically adapts the Canny edge detector and Hough transform to offload compute-intensive tasks while keeping overall latency in check. The main findings show up to 3.7x speedup over a baseline CPU-only configuration, with BOOM cores achieving the best performance and accelerator-enabled designs enabling lower clock frequencies for energy savings. The work demonstrates the practicality of leveraging RISC-V ecosystems (Chipyard, FireSim, AWS) to prototype and evaluate vision workloads for autonomous vehicles, and suggests future exploration of larger-matrix workloads and neural-network inference on Gemmini.

Abstract

In recent years, autonomous vehicles have attracted the attention of many research groups, both in academia and business, including researchers from leading companies such as Google, Uber and Tesla. This type of vehicles are equipped with systems that are subject to very strict requirements, essentially aimed at performing safe operations -- both for potential passengers and pedestrians -- as well as carrying out the processing needed for decision making in real time. In many instances, general-purpose processors alone cannot ensure that these safety, reliability and real-time requirements are met, so it is common to implement heterogeneous systems by including accelerators. This paper explores the acceleration of a line detection application in the autonomous car environment using a heterogeneous system consisting of a general-purpose RISC-V core and a domain-specific accelerator. In particular, the application is analyzed to identify the most computationally intensive parts of the code and it is adapted accordingly for more efficient processing. Furthermore, the code is executed on the aforementioned hardware platform to verify that the execution effectively meets the existing requirements in autonomous vehicles, experiencing a 3.7x speedup with respect to running without accelerator.

Optimization of a Line Detection Algorithm for Autonomous Vehicles on a RISC-V with Accelerator

TL;DR

The paper addresses real-time line-detection for autonomous vehicles under onboard power and latency constraints. It proposes a heterogeneous architecture that pairs a RISC-V core (Rocket or BOOM) with the Gemmini matrix-multiplication accelerator, and systematically adapts the Canny edge detector and Hough transform to offload compute-intensive tasks while keeping overall latency in check. The main findings show up to 3.7x speedup over a baseline CPU-only configuration, with BOOM cores achieving the best performance and accelerator-enabled designs enabling lower clock frequencies for energy savings. The work demonstrates the practicality of leveraging RISC-V ecosystems (Chipyard, FireSim, AWS) to prototype and evaluate vision workloads for autonomous vehicles, and suggests future exploration of larger-matrix workloads and neural-network inference on Gemmini.

Abstract

In recent years, autonomous vehicles have attracted the attention of many research groups, both in academia and business, including researchers from leading companies such as Google, Uber and Tesla. This type of vehicles are equipped with systems that are subject to very strict requirements, essentially aimed at performing safe operations -- both for potential passengers and pedestrians -- as well as carrying out the processing needed for decision making in real time. In many instances, general-purpose processors alone cannot ensure that these safety, reliability and real-time requirements are met, so it is common to implement heterogeneous systems by including accelerators. This paper explores the acceleration of a line detection application in the autonomous car environment using a heterogeneous system consisting of a general-purpose RISC-V core and a domain-specific accelerator. In particular, the application is analyzed to identify the most computationally intensive parts of the code and it is adapted accordingly for more efficient processing. Furthermore, the code is executed on the aforementioned hardware platform to verify that the execution effectively meets the existing requirements in autonomous vehicles, experiencing a 3.7x speedup with respect to running without accelerator.
Paper Structure (23 sections, 5 figures, 6 tables, 3 algorithms)

This paper contains 23 sections, 5 figures, 6 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Basic notions and state of the art
  3. Autonomous vehicles
  4. RISC-V-based development environment
  5. Chipyard.
  6. FireSim.
  7. Amazon Web Services.
  8. Platform design
  9. Processors
  10. The Gemmini Accelerator
  11. Adapting image processing algorithms
  12. Canny Algorithm
  13. Hough Transform
  14. Line Detection
  15. Delivering higher performance
  16. ...and 8 more sections

Figures (5)

  • Figure 1: Integrated sensors on an autonomous vehicle sensors-av.
  • Figure 2: Schematic of the subsystems of an autonomous vehicle.
  • Figure 3: Architecture of our heterogeneous platform gemmini-dac21.
  • Figure 4: Original image with detected lines highlighted in red.
  • Figure 5: Time results when executing Workload 2 on Platforms 1 and 4 at 50MHz and 80MHz, and Workload 3 on Platforms 3 and 6 at 50MHz.