Autonomous Vehicles Simulation

[Penny Bozic]

Reliability Engineering Services (RES) delivers valuable insights to the electronics industry. From battery reliability, product design review to accelerated life testing our experts can solve your challenges.

Ansys Autonomous Vehicle Simulation provides a solution designed specifically to support developing, testing and validating safe automated driving technologies. This autonomous vehicle simulation solution saves significant time and costs versus traditional development and testing methods by allowing you to exercise your AV/ADAS software stack in a closed loop, with sensor-accurate synthetic data in software-in-the-loop or hardware-in-the-loop context with the driving simulator of your choice.

Ansys AVxcelerate offers dedicated simulation features for sensors and headlamp components to speed up and improve the development and testing of ADAS and autonomous systems. With the unique real-time and physics-based simulation capabilities of AVxcelerate, users can confidently test and optimize the performance of their intelligent headlamp and sensor perception.

OEMs and Tiers can rely on proven and trustable digital data to complement actual driving sessions and edge case coverage. Leveraging its unique real-time capability, AVxcelerate allows you to leverage virtual testing in the Software in the loop (SiL) or Hardware in the loop (HiL) context following the progress of your design cycles.

Ansys AVxcelerate provides an open architecture that connects Ansys simulation to any driving simulator and toolchain you choose, like IPG Automotive CarMaker or Carla. You can now recreate virtually any real-world driving conditionto test systems under variable traffic, terrain, weather, and lighting conditions.

Ansys AVxcelerate Autonomy is an end-to-end safety-driven toolchain combining statistics and simulation at scale to perform sensitivity and reliability analysis critical for developing ADAS/AD functions. Extensively reduce cost and time to market by quickly performing verification of designs and providing safety justification with traceable data to achieve L2+/L3 signoff and homologation.

Ansys AVxcelerate Headlamp offers a fully virtual driving lab for testing and validating intelligent lighting systems in a controlled environment, all while remaining connected with control law models.

Join our Autonomous Vehicle Simulation (AVxcelerate) webinar to explore new features for the AVxcelerate Autonomy toolchain, such as radar model encryption for secure sharing, adaptive grid sampling for faster simulations, improved Lidar accuracy, expanded ODD capabilities, and UI enhancements.

Designed and developed to support the iterative process for cutting-edge sensor and headlamp systems, this solution includes dedicated applications for developing ADAS and autonomous systems. Simulation capabilities reduce the time and cost of physical testing.

Develop dynamic and intelligent lighting systems thanks to a C/C++ API or Simulink toolbox, such as cornering light, adaptive front lighting system, dynamic bending light, adaptive driving beam, matrix beam and pixel light.

Assess perception performances by varying vehicle dynamics, driving scenario conditions (such as lighting), or surrounding traffic situations. Perform NCAP rating tests in co-simulation within the driving simulator of your choice.

Inject accurate raw signals simulated by virtual physics-based sensors for perception in the loop testing. Run reproducible virtual drive on test benches thanks to co-simulation with IPG Automotive CarMaker, Carla or any driving simulator of your choice.

The solution offers a real-time pixel beam simulation and control mechanism capable of managing pixel-level beam behavior (such as DMD or microLED systems for example). Efficiently develop, iterate and test your intelligent lighting system control software, including using hardware-in-the-loop simulation, with camera sensor injection.

Ansys simulation solutions and safety analysis enable a testing strategy to be defined that includes detailed and realistic world modeling and scenario generation that incorporates high-fidelity sensor simulations.

Join us for this free webinar which spotlights how Ansys AVxcelerate will help you efficiently and cost-effectively create your ADAS and AV products, expediting their path to market faster than ever. Receive expert tips for implementing your simulation toolchain for ADAS/AVs.

Join this webinar to learn how you can replace 80 percent of road-driving testing using physics-based sensor simulation. Replicate critical test scenarios to validate your algorithms with our camera hardware-in-the-loop (HIL) test solution.

Enter VISTA 2.0: a data-driven system that can simulate complex sensor types and massively interactive scenarios and intersections at scale. With much less data than previous models, the team was able to train autonomous vehicles that could be substantially more robust than those trained on large amounts of real-world data.

The team was able to scale the complexity of the interactive driving tasks for things like overtaking, following, and negotiating, including multiagent scenarios in highly photorealistic environments.

Amini and Wang wrote the paper alongside Zhijian Liu, MIT CSAIL PhD student; Igor Gilitschenski, assistant professor in computer science at the University of Toronto; Wilko Schwarting, AI research scientist and MIT CSAIL PhD '20; Song Han, associate professor at MIT's Department of Electrical Engineering and Computer Science; Sertac Karaman, associate professor of aeronautics and astronautics at MIT; and Daniela Rus, MIT professor and CSAIL director. The researchers presented the work at the IEEE International Conference on Robotics and Automation (ICRA) in Philadelphia.

LinkedIn and 3rd parties use essential and non-essential cookies to provide, secure, analyze and improve our Services, and to show you relevant ads (including professional and job ads) on and off LinkedIn. Learn more in our Cookie Policy.

Although perhaps, it is only fitted to point out that one of the oldest simulations and associated computers is the Antikythera mechanism. This device was developed by the Greeks sometime in the range of 250 - 50 BC for the purpose of predicting the motion of the sun, moon, planets, and stars.

When it comes to autonomous vehicle simulation it is the task of the simulation engineers to accurately map the correct simulation modeling technique to the necessary testing use-case. This process is needed achieve the performance or behavior predicted under the various operational design domains (ODDs). With autonomous vehicles, often the main software systems that benefit from physics based, high resolution graphics simulations; fall under three main categories of perception, motion planning and controls. Although, simulation use-cases and opportunities are not limited to these three main categories. Typically, a validation and verification strategy starts with these main systems at it's core.

One of the more challenging aspects of AV development and the topic which we will focus in this post on, is the perception component. Perception systems involve processing various sensor data such as cameras, and perhaps others such as radar and LIDAR. By fusing these multiple sensor modalities, the sensor data is used to build an understanding of the world around the AV. It could be as simple as identifying the lane lines on the highway, or as complex as understanding that the pedestrian standing at the upcoming intersection is probably waiting for the walk signal and unlikely to just step out in front of the AV.

For the use-case of generating training camera data for perception development, the synthetic data generated in a simulation is often used to augment additional amounts of training data gathered in the real world. It is likely that more realistic, or higher fidelity, synthetic data would probably be more effective at improving the perception quality. If the additional synthetic data, regardless of fidelity level, can somehow be demonstrated to improve the real world performance of the perception system. Such results would increase the level of confidence in validating perception systems.

It is not the first time we encountered such conundrum in the history of computing. Previously, Charles Babbage inventor of the digital programmable computer concept and early computing pioneer alongside Ada Lovelace, encountered it first hand:

One way we ensure statistical realism in our simulated world is by creating realistic conditions for our autonomous driving technology to experience. So, if the Waymo Driver is going through spring showers in Detroit at sunset in SimulationCity, we can recreate raindrops on our sensors and even simulate other minute details such as the dimming light and solar glare.

Just like it's not sufficient to only test in ideal conditions, it is not enough to test how autonomous driving technology reacts to just well-behaved or overly aggressive road users. We want to understand how the Waymo Driver reacts to the full distribution of behaviors it will encounter in the real world.

Suppose we simulate a tailgating scenario at an intersection. We want to understand as many various outcomes as possible and their likelihood to happen to evaluate the Waymo Driver's behavior. If we picked a random tailgating scenario, chances are the tailgater would brake in time. But it's important that we also assess how the Waymo Driver behaves if the tailgater doesn't brake in time, such as when that driver is distracted or inattentive. As we simulate more and more variations of the same scenario, we begin seeing a convergence of the distribution of outcomes between what we observe in simulation and the real world. SimulationCity also enables us to explore rare events, so as to create risky scenarios the Driver has never encountered before, but are still proven to be realistic and very useful.

c80f0f1006