Jan 6, 2020
Qualcomm products mentioned within this post are offered by Qualcomm Technologies, Inc. and/or its subsidiaries.
Modern vehicles now feature more digital and computing technology than ever before. Rapid infusion of advanced technology has brought the automotive industry to a new inflection point, which is changing the industry in real time and shaping how we drive today and in the future. In this blog post we discuss the key trends in automotive industry, the need for next generation automotive solution and how Qualcomm Technologies’ next-generation platform is poised to change the way we drive.
It’s no longer business as usual for the auto industry
Key disruptive trends in the automotive industry can be described in a 4-letter acronym: C.A.S.E
- C: Connected
- A: Autonomous (commonly referred to as Self-Driving Technology)
- S: Shared ownership
- E: Electrification
Working together, these key trends are disrupting the status quo of traditional automotive architecture and methodologies. Automakers are increasingly reliant on technology innovation to address the solutions for the future, perhaps the biggest being Autonomous driving.
Modern cars are now deployed with advanced driver assist solutions (ADAS) with very different scales of feature-sets and complexity. From a feature standpoint, the automotive industry can be broadly classified into three main segments:
- Active Safety - Designed to meet minimum safety requirements from regulatory bodies. Basic features such as lane departure warning and automatic emergency braking, to assist the human driver. Typically maps to SAE Level 1 & SAE Level 2 systems.
- Convenience - Designed to provide enhanced safety and more convenience features to the driver. Assisted automated driving features such as driver monitoring, highway auto-pilot and self-parking, to offer convenience through limited self-driving (e.g., Tesla Autopilot). Typically maps to SAE Level 2+ & SAE Level 3 systems.
- Self-Driving - For Robo-taxis and driverless vehicles, autonomous driving features that perform all safety-critical driving functions and monitor roadway conditions. Typically maps to SAE Level 4 & SAE Level 5 systems.
However, a closer look reveals that there is an explosion of complexity as systems scale up from basic active-safety functions and this is exactly where more technology innovation is needed to make possible better systems.
Explosion of complexity
Most cars today with ADAS have a forward-looking camera, rear camera, and a radar sensor that work together to enable automatic emergency braking. Such a system is limited in feature-set and does not address tough environmental conditions. In contrast, a premium Highway Autopilot system typically has more than 16 perception sensors and needs significantly higher performance in order to understand the 360O environment around the car. This non-linear performance scale (from active safety to self-driving) makes it extremely difficult for automakers to use the solution from active safety for scaling up to self-driving.
Another key development in automotive is the shift of complexity from hardware development toward software development. The amount of software embedded in a car — besides the software used to design, manufacture and provide support for operation and maintenance — is exponentially greater than it was 10 years ago. In fact, a modern car has more software than most of its contemporaries — a phone, a computer or an airplane for that matter.
As performance of systems goes up, the exponential increase in consumed system power does too. This is especially problematic in automotive because these systems need to be deployed in constrained spaces (i.e., a small form factor) and extremely challenging environmental conditions. Thus, high power systems often require complex and expensive liquid cooling solutions, but this makes deployment and mass production a big challenge.
Data & Security
“Data is the new oil” — a simple mantra for the modern autonomous car. Vehicle architectures need to be adapted in order to allow for enhanced data collection, connectivity, and processing-based architectures. This has driven one of the key changes in vehicle backbone architecture and that is the change toward domain aggregation and centralization, the benefit of such an approach is clear — bringing together relevant domain data under a single controller to have enhanced processing and future scalability.
State of the stakeholders
The explosion of complexity and development challenges associated with autonomous functionality are causing the stakeholders to adapt and change their approach to meet the new reality. Let’s review the state of the stakeholders:
Automakers want to offer more differentiated features into their vehicles; however, they are often limited by the affordability of vehicles. Consider the challenges a typical OEM faces:
- Address the mass-volume market where budgets are tight, margins are thin, affordability is important, and regulations mandate active safety systems
- Develop and deploy comfort systems with a comprehensive set of sensors for premium models
- Invest in technical advances to show continuous improvement toward full self-driving
All of this needs to occur in the face of complex and growing software requirements, additional safety implications, shorter time to market, and differentiation from the competition.
While Active-Safety ADAS systems are now being offered as standard — often because of mandatory legislations put up by vehicle safety regulators — these systems are highly constrained on features, performance, power, and cost. As a result, Active-Safety systems have focused on basic functions such as automatic emergency braking, lane-keeping assist, and traffic sign recognition to primarily assist accident avoidance. On the other end of the spectrum, more advanced systems — which include convenience and self-driving features — are too expensive to deploy en masse, thus limiting broad attach rates.
For a Tier 1 supplier, the challenges are even greater. In addition to meeting different segments (active safety, comfort, and self-driving), the Tier 1s are responsible for integrating the full system, solving complex safety problems and in some cases, undertaking liability. Often, Tier 1s must also perform multiple customizations for different OEMs, making their work even harder.
OEMs have traditionally worked on a platform provided by the Tier 1 and were thus shielded from the complexities of adapting code to different processor architectures. But often, Tier 1s also have to frontline the adoption of multiple new compute/SOC platforms to meet different OEM needs. This is a challenge in itself because the silicon technology supplier space is fragmented. For example, some solutions are black boxes, in other words, not customizable/programable by the automaker but meet basic safety mandates. Others, however, offer higher performance and flexibility but need expensive thermal mitigation solutions, making them too costly and complex for mass deployment.
What consumers want
Tesla Autopilot is probably the best contemporary example of an autonomous driving and parking solution available. Consumers are increasingly wanting to try new features and gain more comfort with such in daily driving. Features such as highway autopilot and self-parking are critical to improving road safety, creature comfort, and ride experience and should be the next targets to be deployed on a mass scale.
This need in the automotive industry to bring advanced level of scalability, ease of development and better user experience to mass deployment is a key focus for us. This is exactly where the new Qualcomm Snapdragon Ride Platform comes in.
Introducing the Snapdragon Ride - end-to-end autonomous platform that scales across all segments
For more than a decade, Qualcomm Technologies has been bringing innovative technologies and comprehensive solutions to the automotive industry. Qualcomm Technologies’ leadership position in smart connectivity, infotainment, and cluster systems is built on developing a thorough, full scale, system-based approach of solving the problem versus merely providing a piece of silicon. Now we are ready to repeat the same success in Autonomous driving with the Snapdragon Ride Platform.
The Snapdragon Ride Platform combines hardware, software, open stacks, development kits, tools, and a robust partner ecosystem to help automakers adapt and react to the changing automotive industry and consumer demands around improved safety, convenience, and autonomous driving.
At the heart of the Snapdragon Ride Platform is the hardware, combining an advanced system-on-a chip (SoC) and next-generation accelerators to enable the kind of scalability previously unavailable to the automotive industry.
Snapdragon Ride Application Processor (AP)
- Next-generation neural processing engine
- Multiple dedicated hardware accelerators
- Efficient heterogeneous compute architecture
Snapdragon Ride Autonomous Accelerator
- Massive Neural processing engine array
- Seamless software integration
- Massive data pipes for highly efficient operation
Imagine having a system scale from sub-30 TOPs to 700+ TOPs with worst case power consumption of 130W @ 700 TOPs.
However, just TOPs metrics are no longer enough to describe the performance of these advanced hardware platforms. For instance, next-generation architecture combines the best of data transfer efficiency, compression techniques, and computational architectures to enable a compute platform that has more than 2x of efficiency versus older architectures.
These platforms are designed to deliver best-in-class performance, with low power and thermal profiles specific to the auto category that help facilitate scalability from active safety to convenience to self-driving using a single software platform, with a massive amount of computing power still available for future innovation.
Qualcomm Technologies offers safety-middleware, operating systems, and drivers — all working together to offer the seamless development experience that automakers and suppliers require. Having these optimized components available not only helps lower development costs, but also helps extract every bit of performance from the hardware.
Open driving software stacks
The experience of safety, comfort and autonomous driving is directly linked to the quality of environmental perception and driving stacks that are running on the platform. This is where Qualcomm Technologies’ approach to design high-quality, open and customizable stacks shines.
The Snapdragon Ride Software Stack has three main components
- Localization & Map Fusion
- Perception & Sensor Fusion
- Behavior prediction & Planning
Built using hundreds of person-years of meticulous planning and hard work, the Snapdragon Ride Autonomous Stack offers an open and robust platform that when combined with Tier 1/OEM software, creates one of the most advanced vehicle-perception and driving brains available. An example of the real-world robustness of this stack can be seen in the nighttime driving shot picture — more than 30+ DL algorithms working concurrently on 16+ sensors with advanced ML/DL/RL-based algorithms to create a safety conscious and comfort-filled driving experience.
Development tools & infrastructure
Often, understanding the development of infrastructure is intrinsically important in the development of embedded automotive technology. Whether it is the ability to collect data, data labelling, debugging, map updating and/or performance optimizations or quick in-vehicle testing, each development tool/infrastructure piece plays a critical role in accelerating development, and is more often a much more time-consuming and costly step.
Qualcomm Technologies’ experience with advanced system and development led to a suite of exhaustive tools designed to help automotive customers during every step of their product development process. From a ready-to-go development hardware platform to an extensive set of cloud-based and lab tools designed for production-intent systems, these tools save significant amounts of development cost and time.
Finally, OEMs have access to suppliers who can use these tools to customize the hardware and software in the platform. Qualcomm Technologies brings the suppliers into a single ecosystem, allowing automakers to decide how they want to differentiate their offerings, and then choose a suitable vendor.
With this comprehensive approach, the advantages of Qualcomm Technologies’ platform are lower development costs and shorter development cycles, based on a single software framework and family of SoCs. Automakers now have a viable alternative to making piecemeal choices at each point of development and making them all work harmoniously.
The Snapdragon Ride Platform empowers automotive suppliers and automakers to deploy virtually all the features they want now, with the ability to scale in the future. Our custom-built, auto-ready workload architecture features extreme performance with low power consumption and more simplicity to aid with automotive safety.
Delivering what’s needed, and what’s next.
While the automotive industry has advanced by leaps and bounds in improving automotive safety, consumers will continue to expect features that provide more comfort and safer driving.
Our comprehensive platform with high performance, low power, open software and multi-ECU aggregation capabilities helps Tier 1 suppliers and auto OEMs to scale from active safety to comfort to full self-driving with lower cost of development overall. And our unique, open stack combines an optimized software offering with hardware to provide greater customization and transparency, helping automakers to go to market quickly.
Automotive designers now have a smooth, clear development path from active safety into the promising market segment of convenience features. Plus, as their customers start demanding features for full self-driving, Qualcomm Technologies’ open platform offers them the competitive edge of scalability, lower development costs and shorter time to commercialization.
Follow our upcoming posts in this series for technical insights into the platform. OEMs and automotive designers will see how they can build comfort and convenience into autonomous driving without starting from scratch. Meanwhile, find out more about Qualcomm Technologies’ approach to automotive compute and wireless.