AliOS Cyber smart cab operating system: AliOS Cyber, which began mass production in 2021 for smart cabs and supports a rich ecosystem of smart space computing engines, has the following features: 1. Multi-core heterogeneous distributed architecture: meet the functional isolation and functional safety requirements of different domains in the vehicle such as infotainment and instrumentation. 2. Multi-screen interactive display: break the physical boundary between screens, meet the random operation and circulation of man-machine windows between different screens, and bring unbounded interactive experience 3. Powerful multimedia support capability: support multi-screen multi-sound zone playback and multi-screen synchronous playback of the whole vehicle to create panoramic space audio experience, support simultaneous 10-way BLE Bluetooth device connection to create exclusive second space. 4. Wide ecological compatibility: deeply integrate mainstream cab development ecology such as Unity/Unreal/Kanzi/Mini App/CAF to create a human-computer interaction experience with rich sense of technology and immersion; support Qualcomm 8295, 8155, X9 series, MTK and other mainstream vehicle chip platforms. 5. Service integration: flat service access mechanism, support SOME/IP, DDS multiple protocols, realize atomization of vehicle and cab services.

AliOS Drive intelligent driving operating system: AliOS Drive intelligent driving operating system was officially released in 2022. It is oriented to intelligent driving scenarios. It has core features such as dual-core drive, layered decoupling and cross-domain sharing. It is expected to be mass-produced in 2024. It has the following characteristics: 1. Open and controllable With the technical characteristics of layered decoupling and cross-domain sharing, it meets the demand of Chinese market, and continuously iterates to build an open and controllable technical system for auto enterprises. 2. Dual-core drive Self-developed microkernel AliOS RTOS and AliOS Safety Linux form the dual-core driver. Efficient optimization with safety, performance and development ecosystem 3. High security Both AliOS RTOS and AliOS Drive Hypervisor have passed the highest level of automotive functional safety ASIL-D product certification. 4. High real-time performance Secure microkernel and Hypervisor provide hard real-time guarantee, scheduling delay/interrupt delay <5us; Safety Linux real-Time enhanced, with average scheduling latency <10us 5. Support mainstream chip platforms Support Horizon Journey 5, Black Sesame A1000, X9 series, TI TDA4, NVidia Xavier/Orin, Infineon TC397 and other mainstream vehicle chip platforms 6. Reliable tool chain assurance Self-developed TCL3-level security compiler/tool chain to provide reliable tool guarantee for intelligent driving system and middleware.

Banma Co-Pilot: Based on Tongyi large model, it is built for automobile scene, and builds cloud-integrated full-stack AI capability. It has core capabilities such as scene customization, multi-source data, professional knowledge, service access, behavior prediction, etc., realizing personalized experience and truly opening the era of "user-defined car". Hello Banma-Self Learning 2.0: For complex speech, by integrating the DAMO Academy Tongyi Model, the Banma voice assistant can learn from the Tongyi model in real time and correctly understand the user's intention. When the user says similar expressions again, it can also be directly understood and executed. As for self-learning 1.0, users also need to personally teach the car machine system to learn. But for self-learning 2.0, which integrates large model technology, it can directly learn in real time through large models, and the learning efficiency is greatly improved. Hello Banma-Free Dialogue: Banma voice assistant grasps a large amount of encyclopedic information by integrating the Tongyi large model of DAMO Academy and combining the data accumulation and training of the vehicle environment. Besides, there are better multi-round dialogue, more logical reasoning, and richer generation capabilities, bringing new interactive experiences to users. Variety space: one-sentence scene generation, creating a "variety space" that truly meets the needs of users; for examples, the ability of text-to-graphic generation, upgrading from offline to online, from deep sea space to historical future, with unlimited imagination and possibility. The generation ability is more than just a graph; it involves the model reasoning SOA atomic terms, scheduling spatial changes, and integrating five senses.