Simulation models can tell engineers a lot about their designs. Will an axel buckle under stress? Is the coefficient of drag lower than it was in last year’s model?
But, at the end of the day, simulations can’t tell you what it’s like to be in the driver’s seat.
“It’s hard to have a model with a bunch of numbers tell you if that subjective feel is good or bad,” said Tony Spagnuolo, vice president of Business Development at VI-grade. “Simulators are that one step further to make those models that much more effective.”
Think of it like short-circuiting the design process. Instead of building the models and prototypes, you put the models into a simulator to test drive the vehicle right away. If you think about it, having a middle step between the virtual and physical models, in the form of a simulator, can save a lot of time, travel and prototyping costs. Spagnuolo notes that some of his driving simulator customers have quoted cost and time savings to be around 40 percent.
“We have this tagline: Bridging the Gap between testing and simulation,” explained Spagnuolo. “We want customers to be able to dream it, try it and then make it. The traditional way is you dream it, you go design it, make it, try it and then you iterate. Ultimately, we want a methodology that allows faster, easier and safer development.”
Some of the obvious applications of engineering-based driving simulators include the optimization of:
- Ride and handling
- Noise, vibration and harshness (NVH)
- Advanced driver assistance systems (ADAS) and autonomous vehicles (AVs)
- Human-machine interfaces (HMIs)
- Motorsport vehicles
But how do engineers get from simulation models to a driving simulator?
What Is a Driving Simulator?
First let’s handle the basics.
A driving simulator is a system of visual and physical tools made up to emulate the experience of controlling a vehicle on a road, track or raceway. At the bare minimum, this includes a screen displaying a visual simulation of a vehicle that is operated by physical representations of a control system. In the case of a car, this physical representation can be as simple as wheels and pedals. The most in-depth simulators, however, have details that include moving physical representations of the cabin, haptic feedback systems and simulations based on real-world physics.
For much of the public, driving simulators are rides or games at high-end arcades or amusement parks. With these simulators, someone sits in a cockpit made to look like a race car so they can play out a fantasy.
To engineers and designers, however, a driving simulator can be an essential tool. It enables them to optimize the look and feel of future vehicles while they are little more than code, diagrams, simulations and designs on a company server.
Spagnuolo said, “At the end of the day, you buy a vehicle based on its subjective feel, handling and sounds. Traditionally, you really didn’t know until you had the vehicle itself what the overall experience [would be]. The objective of simulators is to bring those models in and allow a driver to get that complete subjective experience.”
How Do Driving Simulators Work?
From an engineering standpoint, driving simulators run physical models in the background that interact with a user in real time. If the user turns left, the simulation will perform all the calculations needed to understand how that input affects the models within the simulator.
As a result, driving simulators tend to become more complex as engineers move through the development cycle of a vehicle. As each part is modeled, the simulator becomes much more intricate.
Spagnuolo explained that an engineer can try out a model on their desktop, in real time, and then move up the various simulators to ensure that everything is working. Then, once the software graduates to the complexity of a dynamic simulator, real assessments of the vehicles can be performed using test drivers and full-motion haptic feedback.
Alternatively, engineers can use standard vehicle models in a dynamic simulator with a part or two switched out. This helps them assess how those individual parts may perform and affect the overall vehicle. For instance, a company like Goodyear can swap out tire models on a standard vehicle simulation to test product performance.
“Think of our simulators as a systems integration platform,” Spagnuolo noted. “We want people to come with all the bits, controllers, sensors, drivers, tire models, and do the systems integration on the simulator early in the development cycle … for an immersive subjective experience. And then leave the prototype to the stuff you can’t completely emulate and model.”
As for the simulator’s hardware, the most advanced simulators utilize near-360-degree screens to simulate the surroundings, and use actuators and/or cables to move a cabin in a method that emulates the movements of a vehicle.
For instance, the latest simulator family introduced by VI-grade are cable driven. The cable driven simulators offer a dynamic cockpit with nine degrees of freedom. The cabin’s planar movement can be performed using actuators, but they limit the size of the simulator within a given area and can introduce power concerns. As a result, the DiM400 (the first in this cable driven family) consists of two cables, driven by four motors, that control the cockpit’s main planar degrees of freedom. An inertial compensation system (ICS) then isolates actuators from the cables to give the cabin additional, uncoupled degrees of freedom.
How Accurate Are Driving Simulators?
The question shouldn’t be “how accurate is a driving simulator?” Instead, the focus should be on how accurate it needs to be for a given application.
As stated previously, during early development a lot of the tests can be done with a desktop setup. At this stage of the development cycle, the desktop model should give engineers all the information they need to proceed with their work. For instance, an engineer might use a setup like this to verify if a model is working or to test the NVH of a vehicle as it drives in a cityscape.
But, as development moves closer and closer to completion, the simulator needs to become increasingly accurate. For instance, Spagnuolo describes simulators with “a comprehensive steering system that allows automotive realistic steering feedback, along with active steering, which accentuates that feedback to be more accurately driven. So, the driver at the steering wheel gets the actual response they would get from the vehicle.”
Advanced simulators are so accurate that they can be used to help verify and sign off on a vehicle’s design. These simulators can be preprogramed with company and regulatory standard tests that ensure the safety of everyday vehicles.
“With systems like VI-WorldSim, you can put the car in a track or city and include traffic pedestrians, cycles and animals. And you can give that environment a prescribed or random path,” Spagnuolo said. “You can then interact with the traffic, and so on, as you’re driving the vehicle. You can establish scenarios, and you can establish standard ones. There are regulatory standards—some include external factors, and some don’t. I think we will see an expansion of those regulatory standards and the customers will make other ones.”
Simulators can also be used at the automotive sport or research and development level. In much of these scenarios, the simulator will need to become a digital twin of the physical vehicle so that tests can be performed in a safe environment. Spagnuolo explained, “For racing vehicles and autonomous vehicles, there is a need to get more yaw to get that subjective experience. The DiM400 is the most capable system to do this today.”
Are Driving Simulators Expensive?
To an engineering team, the calculation of the expense of a driving simulator can be better answered by assessing the risks of not using one.
Spagnuolo explained that when you produce offline models, it is hard to correlate it to true behavior. But when you take a model and bring it into a simulator, the process is improved as you can test if it provides a certain feel. You get a level of correlation you cannot get other ways. The person that drives the prototype has more sway on the design, so this brings the modeling, development and test people closer together in a collaborative environment.
Through this collaborative environment, the design team can better understand the vehicle-as a whole. This makes it far less likely that unexpected issues will come up during the prototyping stages-or once the car is on the market. At that point of the product life cycle, the expense of these issues could be catastrophic.
“Until you take your model into a simulator, you really don’t know how good it is. That doesn’t mean you are doing a bad job. It just means now you’ll find out how good this thing really is,” Spagnuolo elaborated. “People are amazed by how well it improves the modeling process by having a simulator. And how it gets the development engineers, that historically wouldn’t touch the model and strictly drive the vehicle, embedded with the modeling engineers. This gets the organization together in a collaborative fashion and transforms how you develop the vehicle.”
But budgets are budgets. Many teams cannot afford the use of a fully dynamic simulator. In such cases, companies will provide ways to rent out this equipment as needed. For instance, VI-grade can deliver a complete system that includes:
- Baseplates
- Motion platforms
- Screens
- Projectors
- Software
- Computer IT infrastructure
The result is a fully installed operating simulator.
The Future of Driving Simulators
As cars move to Level 4 and Level 5 autonomy, driving simulators will be less about the subjective feel of the driver and more about the feel of the passenger.
“Nothing stops you from replacing the driver of a simulator with a machine learning algorithm and now putting a passenger in the vehicle and measuring their subjective interpretation of that vehicle,” Spagnuolo said.
“If you buy an autonomous vehicle, you’re going to be the passenger and you’re going to want to subjectively feel like you’re driving a great vehicle. That means it doesn’t jerk you around, have poor ride quality, or have a twitchy steering wheel. Those things will be important to evaluate before those vehicles are released.”
Spagnuolo adds that currently, the autonomous vehicle market isn’t at the stage of testing consumer reactions. But simulator technology can still be vital to the development of this technology. For instance, it could be used to train and test machine learning (ML) and artificial intelligence (AI) algorithms.
“What it will take will be to run our software stack offline. And you would run it much faster than real time,” Spagnuolo said. “You can run millions offline, hundreds online in a simulator and maybe tens that are actually in a vehicle. There are physical time limitations in terms of how many tests you can run in a simulator.”
By running those millions of simulations offline, ML and AI systems can learn to drive various scenarios created within the simulator software. This will test their effectiveness without risking harm to any physical person or asset.