Jeff's MCAD Blogging
Jeffrey Rowe has more than 40 years of experience in all aspects of industrial design, mechanical engineering, and manufacturing. On the publishing side, he has written well over 1,000 articles for CAD, CAM, CAE, and other technical publications, as well as consulting in many capacities in the … More »
Siemens Making Sense of Autonomous Vehicle Sensors Through Digital Simulation – Part 1
March 29th, 2018 by Jeff Rowe
Although they hold much promise, this has not exactly been a stellar time lately for self-driving/autonomous vehicles. As a matter of fact, recent events have cast a dark cloud over them.
Testing them on the road is, of course, essential, but I’ve often wondered if digital simulation could be used more to maximize safety and efficiency with less road testing required.
As it turns out, this very thing, simulation, is finally being performed quite extensively.
This week, Siemens introduced a breakthrough solution for the development of autonomous driving systems as an addition its Simcenter portfolio that minimizes the need for extensive physical prototyping while dramatically reducing the number of logged test miles necessary to demonstrate the safety of autonomous vehicles.
In a nutshell, this computing and simulation platform is aimed at accelerating the validation and verification of autonomous cars.
Siemens PLM Software Driving Simulator
According to the findings of a report issued by the Rand Corporation, autonomous vehicle prototypes would have to be driven hundreds of millions of miles, and in some cases hundreds of billions of miles, over the course of several decades to demonstrate their reliability in terms of fatalities and injuries – an outcome the authors deemed inconsistent with the near-term commercial viability of self-driving cars. For possible solutions to these challenges, the researchers pointed to innovative testing methods such as advanced simulation technologies.
Leveraging advanced, physics-based simulation and innovative sensor data processing technologies, the new Siemens solution is designed to help automakers and their suppliers address this industry challenge with the potential to shave years off the development, verification and validation of self-driving cars.
The new solution integrates autonomous driving technologies from recent Siemens acquisitions Mentor Graphics and TASS International. TASS’ PreScan simulation environment produces highly realistic, physics-based simulated raw sensor data for an unlimited number of potential driving scenarios, traffic situations and other parameters.
The data from PreScan’s simulated LiDAR, radar and camera sensors is then fed into Mentor’s DRS360 platform, where it is fused in real time to create a high-resolution model of the vehicle’s environment and driving conditions. Customers can then leverage the DRS360 platform’s superior perception resolution and high-performance processing to test and refine proprietary algorithms for critical tasks such as object recognition, driving policy and more.
“Automakers are realizing that physical prototypes and road testing alone cannot reproduce the multitude of complex driving scenarios self-driving cars will encounter. In fact, many of the deadliest scenarios are impossible to reproduce, while others are so dangerous to reproduce that ethics preclude pre-testing,” said Dr. Jan Leuridan, senior vice president, Simulation and Test Solutions, Siemens PLM Software. “It is clear that the near-term commercial availability of fully autonomous vehicles is highly dependent on advanced, physics-based simulation technologies, where Siemens is setting the pace for the larger worldwide automotive industry.”
To deliver the most comprehensive and accurate solution possible, Siemens PLM Software is working with many of the world’s leading manufacturers of LiDAR, radar and vision sensing products to develop physics-based, 3D simulated versions of specific sensor modules.
Compatible with the new Siemens simulation ecosystem, the simulated sensors are attuned using detailed design information from sensor suppliers, and validated using real-world measurement data for optimal accuracy. One of the most important sensor partners is Cepton Technologies, a Silicon Valley-based company notable for its long-range, small-footprint LiDAR sensors. Additional sensor partners will be announced later this year.
“Simulation technology is increasingly valuable to developers of automated vehicles as they face mounting pressures to speed development, validation, and performance of their AV solutions,” said Phil Magney, founder and principal for AV researcher VSI Labs. “Siemens now offers simulation solutions for each stage in the development process from sensors, to processors, to sub-systems, to the entire vehicle. Having a greater scope in simulation solutions offers Siemens the ability to play a leading role in the validation and verification of automated vehicle solutions.”
It All Began With Motorcycles
As autonomous vehicles gain momentum in the transportation sector, you may have heard some skeptics vocalize their opinions against self-driving cars. Absolutely, safety is a top priority for the automotive OEMs experimenting with this technology. These firms are evaluating the active safety and autonomous driving functionalities in a virtual environment with driving simulators by putting the human in the virtual loop. What does this mean? Drivers and engineers are now able to experience the vehicle performance in a safe and realistic environment and analyze the behavior through changing the vehicle’s model and settings.
The Siemens team originally developed a motorcycle simulator during the EC Marie Curie research project, MOTORIST, to study human behavior during dangerous traffic scenarios. This later evolved into a driving simulator to evaluate active safety and autonomous driving functions of cars and their interaction with the driver.
The MOTORIST Motorcycle Simulator
Today, the simulator has a modular design that can easily be adapted for a motorcycle or a mock-up car. Some basic features of the simulator include:
The test rig demonstrates the power of combining state of the art hardware and software solutions to provide a realistic driving experience to automotive design engineers responsible for validating active safety and autonomous driving functions together with human factors.
The work that began for simulating motorcycles evolved and has now moved to autonomous vehicles, including automobiles.
Looking To The Future
Like it or not, self-driving cars are a work in progress, and will continue to be for some time in the (distant) future. Since they employ many technologies from radar to machine learning, these systems must work virtually flawlessly for autonomous vehicles to deliver their passengers to their destination intact and not collide with obstacles, including people, along the way.
Autonomous vehicles are supposed to prevent car accidents, but, unfortunately as we have witnessed, they happen anyway. Designing and delivering self-driving vehicles requires exhaustive testing, and up to now, much of the testing of these systems has been done in the field, where most of the work was done in restricted areas where accidents could damage property but not people.
More recently and increasingly, there’s been a big push to test autonomous vehicles in the real world. Given the state of the technology, this may be have been premature, but the alternative—simulation—has been slow in being widely accepted and implemented. The problem is that the complex simulation required is extensive and sophisticated that extensive computational resources and complex software, all of which has been prohibitively expensive.
Things become even more complex when hardware-in-the-loop (HIL) becomes part of the mix. Hopefully, the Siemens simulation platform will make autonomous vehicle simulation less expensive, comprehensive, and more compelling to more organizations involved with this vital future mode of transportation.
Editor’s Note: A continuation of this article as Part 2 will appear next week as the Featured Story of the Week on MCADCafe.