An Integrated Approach Drives an Autonomous Future
At CES in January 2020, Qualcomm Technologies debuted the Qualcomm® Snapdragon Ride™ Platform, the company’s entry into the ADAS and autonomous vehicle market. Specifically, the platform is initially focused at the sweet spot viewed as Level 2+, which incorporates automated highway driving, self-parking and urban driving in stop and go traffic.
Qualcomm Technologies is utilizing the Qualcomm® Snapdragon™ family of automotive system-on-chips (SoCs) built on a scalable and modular high-performance platform with multi core CPUs as well as GPUs, featuring energy-efficient artificial intelligence and computer vision engines. This platform is combined with an ADAS and autonomous vehicle accelerator and software stack to target the complexity of the highway autopilot systems.
The platform is designed to be scalable to address the growing number of sensors and huge amounts of data required to deliver the various stages between Level 1 to Level 4+ driving capabilities. Each level has a larger degree of performance criteria to deal with increasingly complex and unpredictable environments that require significantly more processing, climbing from sub-30 Tera Operations Per Second (TOPS) for L1/L2 applications to more than 700 TOPS for L4+ autopilot systems.
The Snapdragon Ride Platform is notable because it addresses the industry’s requirement to improve safety and driver convenience, but from an integrated solution perspective. At the basic level, it incorporates data from radar and cameras, then utilizes machine learning models to extract fine semantic object properties from the radar signature to enhance performance in complex scenarios.
However, these onboard solutions will only provide line-of-sight data from about 300 meters, which provides a few seconds for reaction. The autonomous stack in the Snapdragon Ride Platform can also process data from C-V2X inputs, providing non line-of-sight intelligence that helps increase the safety capability of the vehicle.
For automotive OEMs, a significant advantage of this package is the platform’s thermal efficiency. Qualcomm Technologies indicated this is at least two times better that competitive offerings, which is an important characteristic since it can have a significant impact on the layout and cost of including it in a vehicle. If the unit is not thermally efficient, it will require liquid cooling. This may result in added costs for piping of coolant to the device, and places additional constraints on where the platform can be located.
In the case of electric vehicles (EV), where coolant is already required, the added benefit of a thermally efficient hardware platform is the reduction of power consumption from the batteries, helping improve the electric vehicle’s driving range. Since range anxiety is still one the most negative factors in consumers’ willingness to convert to an EV, this should be a very positive attribute.
The Snapdragon Ride Platform is scheduled to be released to development partners in the first half of this year, with the anticipation that production vehicles will have it incorporated by early 2023. For an update on the platform and its performance capabilities, as well as how it’s addressing the different levels of automated driving, register for the upcoming Qualcomm Technologies sponsored webinar “Behind the Scenes of a Safe and Automated Human-Like Highway Driving System,” scheduled for June 3, 2020.
Qualcomm Snapdragon Ride and Qualcomm Snapdragon are products of Qualcomm Technologies, Inc. and/or its subsidiaries.
Image courtesy of Qualcomm Technologies
