Gideon Kedem heads Valens’ Automotive Business team. He brings over 30 years of experience in the Semiconductor and EDA industry.
The automotive industry is revving up its innovative engines, with leading automakers spending as much as 11 figures annually (paywall) on research and development in their quest to build higher-performing, more service-oriented vehicles.
Among their most ambitious goals, of course, is the self-driving car. By 2020, OEMs and auto-tech companies had spent $16 billion on autonomous vehicle projects. Forecasters project that within the next decade, one in 10 vehicles will be autonomous.
The systems needed to power these vehicles — including high-res cameras, lidar and radar sensors — are already being implemented in existing models. But, as many industry leaders know, these technologies have been integrated in an ad-hoc way, without a long-term vision incorporating scalability, cost and functional safety.
As more sensors and connected technologies are added in the coming years, new protocols will need to be developed specifically for such applications. These automotive technologies will be critical for the functionality and performance of connected and autonomous vehicles, as well as the growth of automotive companies.
The technologies integrated into new cars range from those associated with enhancing the driving experience — infotainment systems, for example — to safety-critical systems like the advanced driver assistance system (ADAS). While the former can sustain a small number of errors at minimal cost, there is zero margin for error when it comes to safety-related systems.
Safety-critical ADAS and autonomous driving (AD) systems come with a host of unique requirements. High bandwidth is essential to supporting the full range of applications. Long-distance connectivity is a must, as image sensors are typically located at the edge of the car and must be connected to the car’s computer systems, which are located somewhere in the middle. Meanwhile, wiring needs to go through doors, on the roof and every which way in the car. Zero-latency is also critical to detecting and acting on safety-related events. Add a myriad of functional safety and cybersecurity requirements, and the full scale of connected and autonomous vehicles’ needs become clear.
Why Standards Are Vital
While OEMs have been integrating proprietary systems to meet their ADAS needs in recent years, the industry has been stuck at “level two” of autonomous driving for some time. Part of the reason for this is the proprietary technologies themselves.
In order to move closer toward full autonomy, it is crucial that the industry adopt an industry-standard connectivity solution to connect all the cameras, radars and lidars in the vehicle. In practice, this common standard will allow for the interoperability of products, ensure that connectivity solutions are tailor-made for their purpose, enable economies of scale, promote a healthy ecosystem of suppliers — with ample market choice — and ensure the safety of products.
The connectivity standard gaining industry momentum is A-PHY, defined by the MIPI Alliance. The Alliance’s CSI-2 and DSI-2 interfaces are widely used in today’s car models, but until recently they have been extended using proprietary bridge solutions. These solutions are costly and sub-optimal in terms of link quality. Standardization would allow the technology to be embedded in various connected systems (cameras, radar, lidar and central ECUs) and data streams (DSI and CSI) could be extended without the need for bridge solutions. This would result in a seamless design that ensures functional safety, reduces time to market and lowers total system cost.
What Business Leaders Should Know
The automotive industry has been stuck at level two of autonomous driving for quite a long time now, and the focus has always been on improving sensors or engine control units (ECUs), but decision makers have not always paid sufficient attention to the links between them. What good is a high-resolution camera to detect the road signs and pedestrians if it cannot transmit that data — error-free — to an ECU? And the opposite is true as well — what good is a state-of-the-art ECU if it cannot receive quality data from a sensor?
Without standardization, business development leaders run the risk of being dragged down by the lowest common denominator. To be sure, the automotive industry has seen an abundance of innovative progress in recent years. Sensor vendors are forging ahead with the unprecedented cameras, radars and lidars that can see far beyond the human eye. And ECU manufacturers have correspondingly built up their processing power. But the connections between these components have not progressed, relying on proprietary analog technology that is running out of steam.
A-PHY is one technology, among others, that has the potential to change that. Not to get too technical here, but A-PHY compliant chipsets can support 2-16 Gbps over 10-15 meters of cabling, with a road map to 24-48+ Gbps — more than enough to deal with the very challenging bandwidth requirements of current and future sensors technology. These chipsets are also basically immune to electromagnetic noise, ensuring error-free links between components.
Now Is The Time
With the pace of automotive innovation rapidly accelerating, bringing new features and functionalities to today’s cars, the need for automotive standards has never been more urgent than it is now.
Connected and autonomous vehicles represent a paradigm shift in the automotive industry. Cars are no longer just conveyors from point A to point B. They’re communications, information, safety and entertainment systems on wheels.
The definition of the vehicle has changed, and business development leaders supporting an automotive connectivity standard is vital for building this next-generation vehicle.