The latest Insurance Institute for Highway Safety (IIHS) report indicates that large SUVs may not be as safe as previously thought as they underperformed in crash tests. Standardized crash tests provide a vital baseline of vehicle crashworthiness in fixed situations, primarily moderate overlap, small overlap, and side crash, which help in determining overall safety. For car manufacturers, safety is the primary concern with every vehicle design. What these tests don’t evaluate, however, is what could happen if the vehicle were to optimize its angle of impact in the milliseconds before impact—an optimized minimal crash. This could deliver an entirely new level of vehicle safety even if it couldn’t be tested in a standardized way. Such impact minimizations are possible courtesy of a complex mix of software and hardware, including advanced driver assistance systems (ADAS), anti-lock braking systems (ABS), radars, and sensors. 

To make a concept like optimized crashing possible, it would not be economically viable or practical to physically test every possible iteration on every vehicle. While the industry is already migrating to simulation technology to validate individual design components, these models are, at best, stand-alone and, at worst, operate under limited conditions. This presents an array of limitations as cars become increasingly autonomous. As automakers explore new ways to test performance and evaluate autonomous driving virtually—they typically do so primarily to avoid collisions. We should note avoiding collisions is always the primary objective. In some instances, however, collisions are not avoidable. In those situations, the question becomes whether the impact can be optimized to minimize damage and loss of life. 

Optimized impact analysis requires combining autonomous driving emulators with crash assessment emulators and possibly even environmental emulators to evaluate autonomous driving capabilities under the myriad potential situations it may encounter. In cases where a collision cannot be avoided, this system of systems would analyze an angle of impact to minimize (a) human impact and (b) vehicle damage.

Crash Test Simulation

For several years, the auto industry has used digital twin solutions to help speed vehicle design and production. A digital twin is a replica of a physical system created from combinations of engineering, physics, and computer science models. To evaluate more autonomous vehicles (AVs), creating connections between multiple digital twins allows new variables to be considered. Integrating various independent scientifically accurate emulators into a single integrated reference model with feedback loops creates a hybrid twin. 

Hybrid twins enable designers to explore a much wider operating range and evaluate AVs’ multitude of sensing and reaction capabilities. The dynamic model augments the expense of physical prototyping and condition limitations of real-life crash tests when assessing performance and safety. Hybrid twins can run tests faster across a much wider operating range, not just processing more scenarios that an AV may encounter, but also collision avoidance and minimization in complex scenarios too difficult to replicate using physical crash tests. Additionally, this model supports a holistic viewpoint of the vehicle experience beyond basic functionality. The simulation allows engineers to analyze the entire life cycle from the model prototyping to the actual performance on the road and modify the design prior to physically building the car. This ensures that functionality specifications are met and the vehicle performs as expected across a much wider range of circumstances.

The hybrid model allows automakers to test the mechanics, reaction speeds, and crumple zones on the vehicle across a much wider array of angles of impact. This could allow for slight shifts in impact angles to take advantage of areas where the vehicle’s mechanical strength is greatest.  

Vehicle Design Shifts Left

Hybrid twin technology enables manufacturers to optimize design changes earlier. For instance, what if your vehicle’s radar sensor had 20% finer resolution or could detect incoming angular velocities 15% further? Could that design contribution be evaluated in the end vehicle while the radar chipset is still in the design phase? Integrated, dynamic models like these provide a way to evaluate designs much earlier in the development cycle. Small changes to hardware and software can be evaluated earlier to configure triggers for protocol-specific events such as errors, message parameters, or payload values. These changes could greatly contribute to resolving critical safety and performance issues before the vehicles go into production, reducing the likelihood of costly defects and delays. To provide perspective, in 2022, 10 million cars were recalled due to software issues alone, costing car manufacturers half a billion dollars.

The onset and expansion of hybrid twins will significantly speed the development lifecycle, reduce fatalities, and model impacts such as power consumption and emissions, which will ultimately aid the drive toward net zero. The end result is the cost-effective production of better-performing vehicles and safer autonomous driving systems.

Hybrid Twins: The Key to Predicable and Safer Autonomous Driving

The days of testing only a few, fixed scenarios, are in the rearview mirror. Vehicles are now incredibly complex machines that need extensive testing, and hybrid twins offer a digital test track for the auto industry to evaluate the entire driving experience, prioritize safety, and ultimately scale AV development.