Improving Reductive Design With Intelligent Data Logging
Industry Insights

/

May 2, 2022

/

#

Min Read

Improving Reductive Design With Intelligent Data Logging

This is an external post, click the button below to view.
View Post

For a century motorized vehicles mainly consisted of materials like metal, glass, and rubber and they burned carbon fuels. This decade is seeing a radical paradigm shift. Among other factors, software costs, which currently account for 10% of the vehicle Bill of Material, are expected to reach 50% by 2030. A revolution in automobile design and manufacturing is underway.

This series of articles surveys the landscape of this mobility shift and covers factors that will determine the future of the vehicle and the vehicle of the future. Last time we covered Maintaining ADAS With Automotive OTA

Today our topic is …

Reductive Design

An evening at the beach, sitting in the car, looking over the crepuscular ocean. The only thing missing is that perfect song from the radio, but you just can’t find your way through the user interface hierarchy. You are touching, tapping, swiping. In the past, there was a clunky button there. You keep searching.

We are certain that in the end the song, a song, will be found, but obviously, there are other, more serious considerations evoked by this scenario. The choice of features in vehicles is constantly growing, many of them in multifunctional interfaces. This poses questions: Is the driver being helped or distracted? Which function should be in which place? How to optimize it? Is a function being used and in the way that was intended by the designers? Never forget that one of the major goals, not only of reductive design, but of all excellent design, is simplification.

Attempts to find out what drivers desire and what they actually do while driving has a long history. Some things are simple: The Citroen 2CV, a legendary French vehicle, introduced heating as a particular luxury in year 8 of a 40 year production run. Hard to imagine pushback from drivers there. Other functions are much more intricate and the desirability varies, in particular when the question of cost comes in.

Decisions on design today range from what still should be a physical button (a bone of contention among scientists, engineers, journalists, and drivers) to which buttons are placed on the steering wheel to the maximum permissible actuating force. In other words: How hard is a given button to be pushed?

To find out what the public desires from a new vehicle, questionnaires and extensive market research have been used. One problem with this approach is the well-known fact that people say one thing and do another. Years of state-of-the-art market research did not save the Ford Motor Corporation from sinking $250 million (in 1950ies money!) into the Edsel, which was one of the biggest automotive flops of all time and, incidentally, a decent car. Methods certainly have improved since then, and one game-changer, in particular for ergonomics and overall safety, was the advent of increasingly refined simulators.

Simulators, Testing, & Research

With simulators, the path was paved for objective and quantifiable results. The driver is available for the whole time of the test. The environment is standardized, as is the given driving situation, which can be repeated with all the study subjects. It may even be possible to, for example, replace instrument clusters or to project them into a mock-up. Various data can be recorded, including data from sensors directly connected to the driver’s body—a ticklish undertaking with real-world drivers.

The restrictions, on the other hand, are severe. Testing capacities are low, which in turn limits the validity of the results. The fact that the subjects are in a test situation may change their behavior. The recruitment of people is another notorious difficulty. While your young nephew may jump at the opportunity to drive in a simulator, your grandma from Dubuque may not, but both will be out on the road and quite possibly in the same model car.

Simulators have developed from actual car chassis, mounted on hydraulically moved stages and surrounded by multiple projections, to complete computer modeling in 3D, using renderings of cars and anthropomorphic data for the modeling of drivers and passengers. This kind of computer simulation removes limitations by time and recruitment of test subjects, but is itself based on collected data and thus even more steps removed from actual drivers and their subjective, real-world experience and performance. 

Another simulation option is to drive on closed courses. Closed-course testing is imperative for the development of technologies that simply cannot be tested on public roads while in development, like automatic braking or adaptive cruise control. However, precise data from real vehicles and driving, in this case, are combined with high expenses and limited driving situations. 

Real-World Data

But things are changing, rapidly. As a matter of fact, today billions of real-world data points are readily available and provided by modern car electronics and software. Besides events like braking and accelerating, interactions with the car's various interfaces can now be recorded and made available for analysis. Every swipe on a screen or touch of a button has the potential to deliver important information. What is being used, for how long, and in what way? How many mistakes are made when trying to locate a certain function? Is the function found or do drivers just give up? How well does voice control work under driving conditions? What features are often used in unexpected ways?

Surprises may await. In the simulator, test subjects possibly accept lane assist as a given, while in reality many of them switch it off as a nuisance. Would it help if the sensitivity was calibrated? In a survey or even simulation, drivers might decline park assist, while in everyday parking they may find it an extraordinarily convenient feature to have.

It gets even more important with the most advanced features of autonomous driving. Do drivers in fact keep proper attention? It is now possible to sense for how long a driver has their eyes on the road and for how long they don’t. Are drivers being alerted in time? This is a question that can’t be conclusively answered in a simulation. 

For automakers, these examples equate to terabytes of data stemming from vastly different ECUs and transmitted from large fleets of vehicles. This makes for an information stream very different from the data points produced by a simulator. These data points have to be collected, filtered, and transmitted, a challenge the automotive industry didn’t face even a decade ago. 

This is the first time in history we can actually see what drivers do and don’t in their everyday driving, get unprecedented insight into a whole range of fields, and accurate answers to questions. The ability to make use of this information is not a “nice-to-have” feature, but rather, it is crucial to success in the market of the future.

The analysis certainly has to be thorough. As a confounding factor, drivers may highly desire a feature they barely use. Some features may turn out to be nothing but a cost factor for OEMs, some may have to be calibrated, and some are just too complicated to use and thus produce anger instead of advantage.

As a result, this allows manufacturers to declutter the car, find a holistic approach, and have the entire driver and their experience in view. Less distraction for drivers means more safety on the road and for everybody out there. The new approach is particularly valid for electric vehicles, which allow for new space concepts.

The Sibros Solution

Sibros’ award-winning software update and data management platform offers the opportunity to collect all data from ECUs, with precise edge-filtering and maximum compression, resulting in minimal hardware usage and transmission cost. This all can be done in a hardware-agnostic manner and according to the highest, certified safety, cybersecurity, and data protection standards.

Instead of using resources for the collection and transmission of data, OEMs can leverage the opportunities that arise from data analysis. Unattractive features can be changed or omitted and existing features can be adjusted, with the goals of more comfort and convenience, added safety and value, and reduced risk and cost.

Sibros’ Deep Connected Platform enables the following:

  • Selective and precise logging
  • Intelligent edge data filtering 
  • Industry-leading data compression
  • Unified platform with Cloud Portal and APIs
  • Hardware agnostic solution
  • OTA firmware/software updates to every vehicle ECU
  • Remote commands and diagnostics

Sibros utilizes a total system design approach that seamlessly connects every fleet vehicle to the cloud to enable OEMs with a single and reliable source of truth for connected vehicle data. 

Conclusion

The automotive sector’s transition from hardware to software will not simply have to be managed, but proactively addressed. OEMs that are in the driver’s seat, that offer new desirable features without causing aggravation by software glitches, will be the winners of this race. Moreover, functions that remain unused are a cost factor to be cut.

Consumers today know that their vehicle is increasingly defined by software. They expect differentiation between brands through technological innovation, without being overwhelmed by features or confused by interfaces.

To know what drivers really want and do, and how they do it, is imperative for future success. This is not a simple task, because consumers are fickle about their desires and there may be surprises in store. We do not expect the ashtray to have a comeback, but maybe a nice push-button, especially for sentimental beach music? 

There are billions of data points to be collected, many functions to be calibrated, production costs to be cut, driver experiences to be improved, protection and safety to be increased—when information is leveraged the right way. Get your fleet on the right track. Talk to us today.

Max Reinhold
Max Reinhold
Max Reinhold has been in the writing and technical communication space for more than 15 years. He spent the better part of his twenties disassembling motorcycle engines—sometimes he was even able to assemble them back again. Model airplanes make him feel nostalgic. He graduated from Humboldt University of Berlin.