Carmakers have the opportunity to reimagine and redefine the entire automotive experience.
The automotive sector is undergoing immense change. The retirement of the internal combustion engine (ICE) in favor of electrified powertrains and a shift towards autonomy has provided carmakers the opportunity to reimagine and redefine the entire automotive experience: how a car looks, works, and behaves, from the tires up.
But such change isn’t easy. An industry founded on the manufacture of heavy steel boxes powered by rumbling, fossil-fueled engines and manually controlled via a dashboard of hard-wired levers, switches and dials has found itself outpaced, outmoded and out of tune with the expectations and realities of today.
A new generation of car buyers, prioritizing personalization and connectivity over horsepower or external styling, has seen the in-car experience become the battleground in automotive differentiation.
Consumers have seen the prices at the fuel pumps. They’ve seen how easily connected, upgraded, and personalized their smart devices are over the air (OTA), and they’re wondering why they still need to take their car to a dealership to fix a software bug. They want something smarter, safer, cheaper, more sustainable to run and more rewarding to own.
So while powertrain electrification and autonomy remain big news (and is likely to be until the last manually driveable, fossil-fueled vehicle rolls of production lines), there’s an even bigger story unfolding behind the scenes – the move to the software-defined vehicle.
Consider the benefits of aggregating all the disparate, hard-coded electromechanical systems of a car into a single, interconnected backbone of technology in which every parameter, every application, and every mechanism is controlled by lines of code rather than physical mechanics and can therefore be upgraded, fixed, and personalized – potentially over the air.
This is the promise of the software-defined vehicle (SDV). Consumers get the promise of a better in-car experience as mileage increases. Dealers get new revenue opportunities and ways to build brand loyalty and customer relationships over time through new features and enhancements. And carmakers, OEMs and tier one suppliers get the opportunity to fix issues instantly, avoid expensive recalls and battle to deliver a safer, more efficient, and more enjoyable experience than their competitors.
A software-defined vehicle is a future-proof vehicle. Imagine stepping into your car and being informed that its brakes are now 5 percent more efficient or that its battery will now last 5 percent longer. We’re used to this with our smartphones – now it’s time to expect it of our vehicles.
An SDV in which every function is controlled by software is also capable of operating autonomously – so long as the physical sensors required for it to understand its surroundings and environmental conditions are present.
Today, most autonomous capability centers around advanced driver assistance systems (ADAS), such as lane assistance on the highway or adaptive cruise control. This is enabled by a range of sensors, from cameras to mmWave radar, lidar, gyroscopes, and GPS.
All of these sensors generate large amounts of data – and even if a vehicle isn’t actively driving itself, this data is a goldmine in training AI models to drive autonomously – learning from how they are driven in each scenario to devise the optimum output from any given situation and the implications of each decision.
But herein lies a larger problem. Forecasts have suggested that while modern vehicles generate around 25 gigabytes of data every hour, fully autonomous vehicles will exceed this by more than 100 times.
That’s not just a lot of data—that’s a tsunami. Not only would this push global networks to their limit, it would also mandate global cloud data centers to work harder than ever before, increasing their energy expenditure and carbon footprint exponentially.
Instead, vehicles need to be able to make sense of this data themselves, using what we call sensor fusion to combine data sets in real-time and spot patterns. This is fundamentally important for autonomous driving, but it’s also important for things like predictive maintenance, where the car identifies degradation or variation of some kind and can adjust parameters to avoid damage or mitigate potential safety issues.
That’s why the software-defined vehicle needs hardware capable of handling everything from simple safety check up to complex AI processing within the vehicle.
The compute platform within SDVs is complex, given the number of sensors and actuators required throughout the vehicle. A powerful central computer acts as the ‘brain’ of the vehicle, delivering autonomous driving and general vehicle functions.
This central computer is connected to multiple zonal controllers, which serve as hubs for specific groups of functions. One single zonal controller might support the car’s headlights, folding and heated mirrors, door locks, windows, seat controls, and so on, while another might be focused on powertrain management and ensuring optimal load on the vehicle’s battery.
One step down from the zonal controllers, we have potentially hundreds of microcontrollers (MCUs) integrated into electronic control units (ECUs) throughout the vehicle, each supporting single-function operations (one inside the headlights, one controlling the door locks, and so on).
There’s also a major role for fast, accurate real-time microcontrollers. Real-time processors are designed to always meet timing constraints and are generally reserved for mission-critical tasks such as automotive and industrial applications. Because we can rely on them to perform their functions in a given timeframe, they’re ideal for safety and autonomous applications, whereas a general-purpose processor might encounter a delay, leading to a potentially catastrophic outcome.
All in all, that is an incredible amount of silicon under the hood. The vision of the software-defined vehicle will live or die on how well this vastly complex network of hardware is integrated into each vehicle. Today, ongoing disruptions in the electronics supply chain are exacerbated by higher-than-expected demand with less-elastic manufacturing capacity and longer lead times.
Some Tier-1 subsystem suppliers are bringing semiconductor design in-house to further control supply and inventory and to optimize the overall system. To ensure success, they are hiring semiconductor experts but are competing in a limited talent pool, amplifying the need for automated semiconductor design tools.
Automotive chips have been fabricated using older semiconductor manufacturing processes to save costs. The revolution in automotive electronics, however, is driven by high-value features, which require implementation in the most advanced semiconductor manufacturing processes and the latest process nodes due to the needed performance, design size, and power consumption.
Using AI to overcome this escalating design complexity is showing great promise in both productivity and optimization.
The automobile has changed forever. Electrification and AI have enabled improvements in efficiency, performance, and safety. While some drivers may think only about how quickly they will arrive at a destination, increasingly, there’s more focus on what happens along the journey — the in-car experience.
Intelligent vehicles are enabled by semiconductors, data, and decision-making AI, causing automotive system complexity to grow exponentially and requiring more sophisticated design and test technologies. To achieve the necessary performance as well as sustainable design goals and safe operations, these elements can no longer be designed in isolation. A new design collaboration paradigm must be embraced where engineers from different disciplines unite to achieve their common goal.
 |
 |
 |
 |
 |
 |
Leave a Reply