Qualcomm and Mobileye have announced new technologies that mutually aim to help more OEMs and Tier 1s develop autonomous driving and ADAS features. Both products were revealed in presentations held at the 2022 Consumer Electronics Show in Las Vegas, Nevada.
At the convention, Qualcomm revealed Snapdragon Ride Vision – an open, modular platform designed to aid the development of L2 and L3+ autonomous driving systems and features. The latest in Qualcomm’s Snapdragon Ride series of products, Ride Vision is as a flexible solution that can be integrated into a variety of vehicles. Its compatibility ranges from vehicles that feature a NCAP front camera system, to vehicles with more advanced systems that utilize front, rear, and surround view cameras. The platform’s scalability allows for flexible deployment options and facilitates its support for different levels of automation.
At its core, the platform consists of a software stack built on a four-nanometer process technology SoC (system-on-chip). To provide OEMs with flexibility when developing their autonomous solutions, the chip’s modular architecture allows for the integration of their own map crowdsourcing, driver monitoring systems, parking systems, C-V2X technology and localization modules. The SoC is paired with a software stack developed by Arriver, a subsidiary of Veoneer, to support this flexibility alongside a number of compute functions that work to enhance driver safety.
Arriver’s Vision stack provides OEMs with several customizable features – including the performance power required to enhance the vehicle’s perception capabilities. These enhancements would allow for the more precise detection of static road geometry and dynamic objects, as well as the recognition of traffic signs for global regulation requirements. This functionality is based on a series of eight-megapixel wide-angle cameras developed with custom neural network architectures. Another is a specialized toolkit for the development of the vehicle’s electronic control unit, which has been designed as a reference system for scaling from L2 to L3+ autonomous applications. The toolkit includes a safety OS, hypervisors, and middleware for ADAS and AI with Neural Architecture Search.
While Snapdragon Ride Vision is not expected to be available for vehicle production until 2024, Qualcomm has released its corresponding software development kit. With it, OEMs can flexibly develop and integrate their own drive policies and components ahead of the system’s release.
CES 2022 also saw the announcement of EyeQ Ultra, a new SoC from Mobileye. With it, Mobileye intends to make the development of consumer autonomous vehicles more accessible. The Intel subsidiary has said that the SoC has been purpose-built for autonomous driving and is capable of performing 176 trillion operations per second (TOPS). It leverages five-nanometer processing technology to carry out these operations and independently handle the processing requirements for L4 autonomous driving functions. This capability is carried out by an array of proprietary accelerators, with each one designated to aid a unique autonomous driving task. These accelerators are paired with additional CPU cores, ISPs, and GPUs to process data from the vehicle’s camera and LiDAR sensors alongside its high-definition map and driving policy software.
EyeQ Ultra will support integration with Mobileye’s Road Experience Management technology. The platform uses road data to develop and update road information in real-time. While Mobileye expects the first silicon for the SoC to roll out at the end of 2023, full production is not expected to commence until 2025.