Mediatek has announced four new SoCs (systems-on-chip) within its Dimensity Auto Cockpit portfolio, offering AI in-cabin experiences for next generation vehicles.
Ten months on from the announcement that MediaTek and NVIDIA would partner to deliver a complete range of in-vehicle AI cabin solutions for the next generation of software defined vehicles, its Auto Cockpit CX-1, CY-1, CM-1 and CV-1 were announced at NVIDIA GTC.
Each SoC will come with support for NVIDEA DRIVE OS enabling its use across all tiers of vehicle from premium (CX-1) to entry level (CV-1). Integrating ARM’s v9-A system and NVIDIA’s next-gen GPU accelerated AI Computing, the AI-enabled Dimensity Auto Cockpit platform runs large language models (LLMs) in the car, allowing vehicles to support chatbots, deliver content to multiple displays, provide driver alert detection and other AI-based safety and entertainment features.
In addition to driving aids, passengers can also stream movies and play video games owing to the integration of NVIDIA RTX GPU and graphics with AI upscaling and frame generation. Video conferencing features including AI gaze correction and AI-based audio enhancement are also available.
In DATAportl’s opinion, bringing AI capabilities into the vehicle makes a lot of sense and is a common trend across semiconductor manufacturers and vehicle OEMs in 2024. Enhancing security features, alerts and reaction time to critical issues are obvious benefits but allowing for greater centralization of compute and processing can help to optimize and simplify servicing, software updates and add to the overall consumer experience throughout vehicle ownership.
Whilst we eagerly await the days in the (hopefully not too distant) future when vehicles fully operate autonomously, why not enhance the in-cabin experience for drivers and passengers via the overlay of emerging technology? Support in the vehicle from a “virtual passenger” type assistant that can intercept and read aloud relevant messages whilst you are driving so you can divert and pick up the item missing from that evening’s dinner on the way home, or send an alert to your colleagues that you’re running ten minutes late for a scheduled meeting are examples that could make our lives easier (what AI is being developed for after all).
It strikes me, however, that whilst these enhanced safety, driver and passenger experiences may help sell the next generation of vehicles for OEMs, it’s also likely that the capabilities will form part of the framework for vehicle-to-vehicle communication and information sharing required to enable autonomy in the future. My concern, therefore, is how many emerging automotive systems and platforms will be interoperable and capable of this type of collaboration.It seems MediaTek’s approach is a sensible one with a collaborative thought process in mind. As a leader in connected infotainment in the home, it’s more than feasible to aim to bring a similar experience to the vehicle. Its collaboration means it can take advantage of NVIDIA’s strength in AI, GPU, and ADAS whilst focusing its efforts on cockpit experience and compute. DATAportl expects strong competition from suppliers including Intel and Qualcomm but with more than 300 million passenger cars forecast to ship over the next five years there’s a large opportunity at stake.
