TR-007: Compute Scenarios for Benchmarking Machine Learning for Automated and Assisted Driving Systems

The letters AVCC made into a car shape in purple and orange to make AVCC logo

TR-007 provides essential guidance for Original Equipment Manufacturers (OEMs), Tier 1 suppliers, and stakeholders in the automotive ecosystem.

This is a significant milestone for the AVCC Benchmarks Working Group and we’re proud to be publishing these recommendations.”

— Kasper Mecklenburg, Chair of the Benchmarks Working Group

SAN JOSE, CA, UNITED STATES, June 4, 2024 /EINPresswire.com/ — The Autonomous Vehicle Computing Consortium (AVCC) today announced the publication of Technical Report 007 (TR-007) “Compute Scenarios for Benchmarking Machine Learning for Automated and Assisted Driving Systems.” Developed collaboratively by industry experts, TR-007 provides essential guidance for Original Equipment Manufacturers (OEMs), Tier 1 suppliers, and stakeholders in the automotive ecosystem.

TR-007 addresses the critical need for standardized benchmarks in evaluating machine learning (ML) models for autonomous and assisted driving. By breaking down complex driving functions into manageable building blocks, the report defines four compute scenarios that stress test systems effectively. These scenarios enable fair comparisons and informed investment decisions, driving innovation and fostering cost-effective solutions.

“This is a significant milestone for the AVCC Benchmarks Working Group and we’re proud to be publishing these recommendations,” said Kasper Mecklenburg, Chair of the Benchmarks Working Group and Principal Autonomous Driving Solution Engineer, Automotive Line of Business, Arm. “As a next step, AVCC and MLCommons are collaborating on the development of the MLPerf Automotive benchmark suite, with the aim of uniting AI benchmarking efforts for automotive and creating the industry standard AI benchmark for the sector.”

Key features of the report include:

– Quantization of data produced in the different processing stages

– Example vehicle compute architecture

– Defined compute scenarios for effective system stress testing

– Framework for informed investment decisions in automotive technology

The release of TR-007 marks a significant step for the automotive industry, enabling stakeholders to accelerate development and deliver safer and more advanced products to consumers.

For more information or to view a complimentary abstract of AVCC’s TR-007: Compute Scenarios for Benchmarking Machine Learning for Automated and Assisted Driving Systems, please visit AVCC’s website at https://avcc.org/tr007/.

About AVCC

AVCC is a global automated and autonomous vehicle (AV) consortium that specifies and benchmarks solutions for AV computing, cybersecurity, functional safety, and building block interconnects. AVCC is a not-for-profit membership organization building an ecosystem of OEMs, automotive suppliers, and semiconductor and software suppliers in the automotive industry. The consortium addresses the complexity of the intelligent-vehicle software-defined automotive environment and promotes member-driven dialogue within technical working groups to address non-differentiable common challenges. AVCC is committed to driving the evolution of autonomous and automated solutions up to L5 performance. https://avcc.org/

Sarah LaLiberte
AVCC
+1 978-502-8558
email us here
Visit us on social media:
X
LinkedIn