Automotive companies need to evolve their development processes to adapt to the new age of software on wheels.
The automotive market is facing tough challenges when it comes to recalls. It seems that every month (if not week) brings a long list of new recalls. It's not that recalls did not exist in the past, but it seems that their numbers, types, and obvious visibility through the media have significantly increased.
We, of course, have all heard about the Toyota bit flip as well as the recent GM recalls in the media, but details about recalls are also delivered to us directly in our mailboxes.
As an example, I recently received a recall for my Toyota Prius (a car I like, independently of this recall).
The envelope stated: "Software Update for Motor Generator ECU and Power Management ECU." The problem was described as follows:
Inside the Hybrid Inverter Assembly is an Intelligent Power Module (IPM) which contains a control board equipped with transistors. Certain transistors could become damaged when operating the vehicle under high-load driving conditions. If this occurs, various warning lamps on the instrument panel will illuminate. The vehicle should enter a fail-safe mode limiting the vehicle's power, allowing it to be driven a short distance. In some cases, the motor/generator ECU could reset, causing the hybrid system to shut down, resulting in the vehicle stopping while being driven and increasing the risk of a crash.
As I was reading the details of the recall, my immediate reaction was, naturally, to first figure out when I could bring my car in for the software upgrade.
My second reaction, having been involved in automotive and software development for many years, was more focused on what could be done to prevent such a recall in the future.
Virtual prototyping is clearly an approach that makes sense when it comes to software development, integration, and test. The concept of virtual prototyping is relatively simple: Create a simulation of your electronic control system including the electronic hardware (microcontroller) and the mechanical system it controls (some refer to this environment as a virtual hardware-in-the-loop environment). Simulating such a system allows developers to start development, integration, and test earlier; test corner cases without risks; spend more time on testing; perform fault testing, etc.
This approach, as any new design process change, requires some level of investment and commitment from companies deciding to deploy it. The benefits and experiences using virtual prototyping have actually been documented. You can find companies like Bosch, GM, and Hitachi Automotive Systems sharing their use cases in an e-book called Better Software. Faster!
So why aren't more companies taking steps to implement the use of virtual prototypes?
Of course, there are some that will say the technology is not fully mature yet, but an increase in documented usage will prove them wrong. Even if some deployment challenges may still exist, I believe that the industry is actively working to address them and bring further benefit to companies deploying virtual prototyping more widely.
So if not a technology issue, what is slowing companies down? As with any design and development methodology, the return on investment could be in question.
However, the ROI gained from using virtual prototyping has been demonstrated both quantitatively and qualitatively. So is it possible that the investment required was simply not manageable, presenting a barrier too high to pass for automotive companies?
As I pondered this question, I recalled a recent discussion with a leading OEM company faced with a recall.
The engineering management explained at great length that their project was going to be reduced in scope (and some parts canceled entirely) because the recall was going to cost the company a significant amount in legal and settlement fees.
And here it was, right in front of me -- the chicken-and-egg problem. To prevent recalls, companies need to invest, but investing means money needs to be available; and unfortunately this money was used to address the recent recalls! So the investment in virtual prototyping could not happen, while this was clearly a possible solution that could prevent more of these recalls in the future.
While I have no doubt that design negligence causing harm to others should be duly compensated, I also believe that automotive companies need to evolve their development processes to adapt to the new age of software on wheels.
If they expect safe cars to be on the road (autonomous driving will become a reality), automotive companies and government regulators need to allow for investing an equal amount (if not more) in protecting the people affected by past negligence and prevent recalls from happening in the first place. This requires investment: The right balance must be found between legal fees and enabling innovation in the automotive design processes to improve safety for generations to come.
— Marc Serughetti is Director of Business Development for System-Level Solutions at Synopsys. He is responsible for driving the development and deployment of virtual prototyping technologies.