Customer solution: Autonomous racing with HiL computer

3 times faster development thanks to HiL simulation

Customer

TU Munich

Students

> 50.000

Employees

> 12.000

Challenges

The race car from the Technical University of Munich (TUM) for the Abu Dhabi Autonomous Racing League (A2RL) drives and controls the vehicle autonomously over the entire racetrack. Developing an autonomous driving system is a significant challenge. To safely and accurately drive the vehicle around the racetrack, real-world driving situations must be tested in a safe and efficient environment. Traditional test methods that require the use of road vehicles are often difficult and costly to implement. Therefore, virtual simulations are used to test the software’s response to real-world scenarios.

However, these simulations require powerful CPU and GPU performance. The large number of different scenarios to be tested – including GPS failures, extreme heat, and other influences – requires continuous and repeated execution with slightly changing parameters. The challenge therefore is to provide and optimize a powerful computer to run reliable and accurate simulations quickly, and ideally, to run multiple simulations simultaneously.

Source: Software architecture of the TUM Autonomous Motorsport team. LiDAR, Light Detection and Ranging; TUM, Technical University of Munich.

Solution

TUM uses the InoNet HiL computer Mayflower^®-B17-LiQuid to simulate the computer installed in the vehicle and to test different scenarios with its software. This system is a premium solution that provides the computing power needed to run complex autonomous driving simulation scenarios thanks to the large number of cores and high frequency of the EPYC 7313P CPU. In addition, the very large memory of the water-cooled RTX A6000 GPU allows the software to run multiple simulations simultaneously to address a wide range of factors.

This means that even minimal delays of a few milliseconds can be optimized for faster vehicle response. In addition, the QuickTray^®-v3 with hot-plug capability makes it easy to evaluate simulation results after the fact.

Case Study Racing mit HiL Rechner Simulation

Customer benefits

The use of the HiL computer based on the Mayflower B17 platform offers the Technical University of Munich (TUM) significant advantages in the development of autonomous driving systems. By simulating driving situations, the HiL computer enables more cost-effective and up to three times faster development by replacing realistic road tests with virtual ones. This not only increases safety by allowing dangerous scenarios to be tested safely, but also increases accuracy thanks to the computer’s high processing power and memory capacity.

The platform also offers increased flexibility and scalability for different test requirements and growing needs. Thanks to its user-friendly operation, it can be used without any special previous knowledge. In addition, the Mayflower^®-B17-LiQuid enables simultaneous simulation of multiple hardware components, simplified deployment of software containers, and optimized container mapping for improved performance. By optimizing predictive capabilities, the system also contributes to more effective decision making, further increasing the efficiency and safety of autonomous vehicles.

Customer feedback

“We at TUM Autonomous Motorsport have the challenge of constantly monitoring and optimising the computing time of our software on our vehicle PC.
InoNet enables us to do just that with its hardware platform. We particularly appreciate the fast, uncomplicated and, above all, highly professional cooperation.” (S. Sagmeister, Teamlead TUM Autonomous Motorsport, Chair of Automotive Engineering)

TUM’s victory at the A2RL

On 27.04.2024, the Technical University of Munich won the race at the Abu Dhabi Autonomous Racing League! The autonomous software developed and tested with the Mayflower^®-B17-LiQuid HiL computer beat the competition to take first place in the Abu Dhabi Autonomous Racing League.

Congratulations to the entire TUM Autonomous Motorsport Team for this outstanding achievement!

Source: Technical University of Munich

Source: Technical University of Munich

Related products

InoHiL®-B17QW-SP3

19“ liquid-cooled HiL server for complex model calculations, signal processing, machine learning and AI

To the product page

InoNet QuickTray^®-v3

High-speed hot-plug & encryption-capable storage insert for data exchange