Chinese EV models. It involves fast and seamless integration of multiple technologies. (Jointly developed by Volkswagen China Technology Co., Ltd. (VCTC), CARIAD China, and XPeng Motors) ADAS It is compatible with Level 2++ advanced driver assistance systems. The AI assist is based on large scale language models (LLM). The system was developed over 18 months by CARIZON, a JV between CARIAD and Horizon, a VW Group-owned software company. (Cre...

Market & Tech Reports  Updated at: 2025/06/11

ojected far ahead, which means that accommodation is easier and drivers only need to slightly lower their normal view of the road.   Interaction between the displays and functions is controlled by BMW’s Operating System X. This also provides the Intelligent Personal Assistant, which uses Large Language Models (LLMs) and enables natural communication with the vehicle to take place – initially for improved navigation functions. The system is...

Market & Tech Reports  Updated at: 2025/06/10

fe Agent, Encyclopedia Chat Agent, Digital Leisure Agent, Local Life Agent, Follow-up Photography Agent, Mission Assist Agent, Communications Office Agent, and Car Service Agent. The System Agent, the main entry point for interaction, can enable scenario-based services for 10 Agents through natural language. It can meet user needs in a variety of situations, including vehicle use, navigation, and family life.   Banma AI Box   Banma AI Box, join...

Market & Tech Reports  Updated at: 2025/05/28

led its shooting brake concept car "EARTH" for the first time. Combining nature and futuristic technology, it expresses the avant-garde mindset of a high-end Chinese luxury brand. The five-passenger luxury smart SUV targeting families, the Hyptec HL, was also unveiled at this motor show. The design language and manufacturing concept of the Hyptec HL and Hyptec EARTH is "inheritance”. Hyptec has created a new "Shanhai Architecture" that is a ser...

Market & Tech Reports  Updated at: 2025/05/20

already provided Deeproute.ai with a wealth of data, which the company uses to enhance its offering. Given their keen interest in technical advancements, Chinese citizens appear to be less critical when it comes to data sharing.   Mr. Xuan went on to say that in China, information based on natural language is produced within the vehicle, especially for the HMI. This helps create predictions and makes it possible to alert drivers ahead of time, i...

Market & Tech Reports  Updated at: 2025/05/16

sensors including NVIDIA's "Orin X" chip, one millimeter-wave radar, and a triple-lens front camera. The vehicle is equipped with the "DiLink 150" smart connected system, a 10.25-inch LCD instrument panel, a 15.6-inch center display and a windshield head-up display (W-HUD). It also features a smart language voice system that utilizes the DeepSeek-R1 large-scale model.   The model is also equipped with other practical features such as 10-point ma...

Market & Tech Reports  Updated at: 2025/05/14

is also being conducted on airbags built into the seats or seatbelts to accommodate the various occupant positions that may occur during autonomous driving.   In addition, the U.S. occupant protection standards, which used to assume the presence of a driver and steering wheel, have been revised to language that does not presume these things, and other changes have been implemented in safety standards and regulations that affect airbag specificat...

Market & Tech Reports  Updated at: 2025/04/24

kers, including BYD, Geely, Chery, and GAC, have released advanced intelligent driving technology solutions and are promoting their installation in all series models. In addition, many automakers, including SAIC, FAW, Dongfeng, and Changan, are using artificial intelligence startup DeepSeek's large language models.   Global automakers are accelerating the introduction of NEV models, including REEVs (Range Extended Electric Vehicles). Volkswagen,...

Market & Tech Reports  Updated at: 2025/04/24

to one hundredth of the power of comparable GPUs. Ambarella demonstrated their GenAI SoC N1-665 in a driving trial in Las Vegas. This SoC enables twelve individual 1080p30 video data streams to be decoded and simultaneously processed on a chip while executing a hybrid of multiple multimodal Vision Language Models (VLMs) and traditional Convolutional Neural Networks (CNNs).   “The high AI-processing performance of these SoCs supports the major...

Market & Tech Reports  Updated at: 2025/03/21

e, please refer to the report 2024 China Auto Software Conference: CAAM, Changan Automobile. The Changan Tops Model is a native end-to-end architecture with Embodied Intelligence. It adopts an integrated model architecture. The model integrates the detection network, planning and control network, language network, and Driving Safety Alignment Reinforcement (DSAR). The introduction of multi-modal large-scale language models enables smart driving...

Market & Tech Reports  Updated at: 2025/03/12

arge-scale models, and vehicle-road-cloud collaboration. These technologies are applied to the intelligent driving experience from departure to arrival. In 2025, it will achieve end-to-end capabilities and have the ability for human-like decision-making. At the same time, the generative VLM (Vision-Language Model) will assist decision-making, significantly improving safety in emergency and complex situations. In 2027, the model will be upgraded t...

Market & Tech Reports  Updated at: 2025/03/07

ies AG (Infineon) and Rohm Co., Ltd. (Rohm) have released 1,700V devices. On the other hand, some Chinese automakers are also developing 1,000V platforms and it is expected that 1,200V platforms will gradually become more popular in the future. Motor design technology based on AI/data large-scale language models Optimization solutions are designed by analyzing large amounts of data and simulation test results. By monitoring and controlling the ...

Market & Tech Reports  Updated at: 2025/02/18

tional software-defined vehicles (SDV 1.0) to “SDV 2.0”, which is centered on AI infrastructure. This evolution means moving from static, fixed, code-based systems to next-generation systems that are data-driven and dynamically responsive. Transformer Architecture is the foundation of the Large Language Model (LLM), also often referred to as the OS of generative AI.   Mr. Wei explained that in the automotive industry, the mainstream approach...

Market & Tech Reports  Updated at: 2025/02/12

vides companies a way to differentiate between brands or models on the same software and allows designers to make aesthetic changes without involving developers. The demonstration unit also showcased a generative-AI powered customization system, which provides cockpit customization based on natural language voice prompts, enabling a new level of vehicle personalization for the end user. For example, if a user asks the cockpit to create a theme ba...

Market & Tech Reports  Updated at: 2025/02/06

Real Estate Data Control Regulatory Compliance Affordable Competitive Solutions AI Race Strategic Control Accelerated Innovation Cycle Reliable, Cost-effective Scalability Digital Monetization Strategies   Up to this point, Cerence has developed the Cerence Drive, equipped with natural language and core domains; the Cerence Assistance, with full stack assistant, wide domain coverage, and Scripted UX; and finally, the Cerence Co-Pilot...

Market & Tech Reports  Updated at: 2025/01/30

cockpit has a panoramic screen spanning the cabin’s width, while two 12.9-inch screens for rear passengers are an optional feature for Signature trims. All displays use the Unreal Engine for visualizations. The Afeela Personal Agent is a virtual assistant which uses AI to interpret conversational language and allows occupants to control specific vehicle features, improve personalization, and receive suggestions. Active noise-cancelling technolo...

Market & Tech Reports  Updated at: 2025/01/24

ity of the huge amounts of data that are required to train the algorithms.   About ten years ago, the first AI systems achieved a level of maturity that enabled them not only to beat humans in a game of chess, but also, for example, to be better than humans in recognizing images and in reading and language comprehension. In 2023, AI systems were already close to surpassing the potential of humans in making logical predictions based on historical...

Market & Tech Reports  Updated at: 2025/01/22

a support capabilities through smart search and long-tail data synthesis, among other measures. Features Details Smart search Traditionally, they have relied on manual labeling methods, which are inefficient and difficult to handle for complex scenes. The current method combines a large-scale language model to enable cross-modal searches, such as text-based photo search, photo-based photo search, and text-based video search, to quickly extr...

Market & Tech Reports  Updated at: 2025/01/08

passenger seat, to project images, providing a rich visual experience for the users.        L5 - Vehicle Intelligent Application Layer: Mainly composed of the Tianshu large-scale model and the Changan intelligent driving system. The Tenshu large-scale model introduces a multi-modal large-scale language model that contains a wealth of knowledge about the world. Intelligent driving systems make it possible to simulate how people interact with ...

Market & Tech Reports  Updated at: 2025/01/08

d control] (Source: Created by MarkLines based on the presentation content)Note:<>indicates AI model, [] indicates rule-basedNOA=Navigate on Autopilot, APA=Automatic Parking Assist, HPA=Home-zone Parking Assist, AVP= Automated Valet Parking, BEV=Bird's Eye View, OCC=Occupancy Network, VLM=Vision-Language Model   ZDrive Product plan Domestic (China) product plan High-speed NOA in 2024-2025, end-to-end Urban NOA without map in 2026-2027, and...

Market & Tech Reports  Updated at: 2025/01/07