The new Snapdragon Automotive Cockpit Platforms are the first-announced scalable AI-based platforms claims Qualcomm, and it notes that the platform is designed to "transform in-vehicle experiences." Those experiences will include in-car virtual assistance, natural language interactions between driver and vehicle, and contextual safety use cases. The system supports infotainment experiences for both drivers and passengers. The platform leverages the Snapdragon 820a platform with immersive graphics, multimedia, computer vision, and AI capability. Integrated heterogenous computing capabilities of the platform include leveraging the multi-core Qualcomm AI Engine, Qualcomm Spectra Image Signal Processor (ISP), fourth-generation Qualcomm Kryo Central Processing Unit (CPU), Qualcomm Hexagon Processor and sixth-generation Qualcomm Adreno Visual Subsystem.
The platforms support multiple operating systems including Android and Linux among others. Advanced wireless tech is integrated including multi-mode cellular connectivity, WiFi 6, and enhanced Bluetooth. Qualcomm is demoing the third-gen Snapdragon Automotive Cockpit Platform now. While several manufacturers are onboard for the platform, there is no indication of when the first products might come to market.
Qualcomm is also talking up its in-vehicle capabilities at CES 2019, including integration of Amazon Alexa voice assistant. The platform leverages the language processing and speech recognition capability that Amazon Alexa offers. This integration allows voice-based experiences inside the car like in-car virtual assistance and natural interactions between the vehicle and driver. An in-vehicle demo at CES 2019 features Amazon Music, Prime Video, Fire TV, and Audible services.
The demo utilizes the Qualcomm Smart Audio Platform to give immersive natural language instructions including navigation, points of interest, and multimedia services. Qualcomm's Smart Audio Platform provided the basis for a new Alexa Voice Service (AVS) development kit that launched last year that allowed always-ready wake-word detection for responsive and precise voice activation.