Mentor provides roadmap to level five autonomous driving

Andrew Patterson, Mentor's director of automotive business development

Mentor hopes to help tier ones and car makers met their autonomous driving goals with a platform that captures, fuses and uses raw data in real time from a wide range of sensors.

The company, which was known as Mentor Graphics until its recent takeover by Siemens but is now called just Mentor, has been working on the DRS360 platform for two years.

“This is a new direction for Mentor,” said Andrew Patterson, director of automotive business development. “There has been a lot of hype on autonomous driving. We decided two years ago to start a project to develop an autonomous driving platform.”

The platform is a combination of hardware and software, with most of the hardware coming from XS Embedded, which Mentor took over in 2014.

The platform takes input from various sensors including radar, lidar and vision. Rather than processing the individual data streams, which can cause latency and data integrity problems, it takes the unfiltered raw data to provide one image of everything that is happening.

“This means we can make the performance faster, more accurate with all the data available and we have a very low power envelope,” said Patterson.

The unfiltered information from all system sensors goes to a central processing unit, where raw sensor data are fused in real time at all levels. Eliminating pre-processing microcontrollers from all system sensor nodes enables a broad array of advantages, including real-time performance, significant reductions in system cost and complexity, and access to all captured sensor data for the highest resolution model of the vehicle’s environment and driving conditions.

The platform’s streamlined data transport architecture further lowers system latency by reducing physical bus structures, hardware interfaces and complex, time-triggered Ethernet backbones. This architecture also enables situation-adaptive redundancy and dynamic resolution by using centralised, unfiltered sensor data to enhance accuracy and reliability. The signal processing software, algorithms and compute-optimised neural networks for machine learning run on a seamlessly integrated, automotive-grade platform.

The platform is engineered for production to meet the safety, cost, power, thermal and emissions requirements for deployment in ISO 26262 Asil D-compliant systems. The raw data are initially processed in an FPGA – a Xilinx Zynq UltraScale+ MPSoC – in partnership with SoCs and safety controllers based on either x86 or Arm architectures.

“Mentor Automotive’s approach to centralised, real-time raw sensor data fusion represents a new innovation for automated driving system developers,” said Arun Iyengar, Xilinx vice president. “With its extreme flexibility, efficient power operation and highly optimised signal processing capabilities, the Zynq UltraScale+ MPSoC family targets these types of applications and plays a key role in enabling the capturing, pre-processing and fusion of data from a wide variety of sensors.”

Though Mentor has a roadmap that will allow the platform to scale up to full level five autonomous driving, Patterson said the immediate targets were levels three and four.

“We are encouraging tier ones and OEMs to use their own algorithms,” he said. “The platform is intended to be a starting point to help them to market quickly. It is a reference platform for tier ones and OEMs.”

He said live trial would be run during the second half of this year in collaboration with a number of tier ones and OEMs. Though he could not name any of them, he said they included German, North American and Asian companies.

“This has been designed with production in mind,” said Patterson. “It uses automotive design rules and is certifiable up to Asil D.”


Recommend to Colleagues: 

Add new comment

Plain text

  • Allowed HTML tags: <a> <em> <p> <h1> <h2> <h3> <code> <ul> <ol> <li> <dl> <dt> <dd> <strong>
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

Follow Us:

Twitter icon
Facebook icon
RSS icon