Shanghai (Gasgoo)- On January 8, autonomous driving tech company Nullmax announced a collaboration with Black Sesame Technologies to release a multimodal AI large model-enabled smart driving solution based on the latest generation A2000 chip from the Huashan series. This innovative end-to-end architecture and high-performance autonomous driving SoC (System on a Chip) is designed to power next-generation autonomous driving applications across various scenarios.
Debuting globally at the CES 2025, the aforesaid new smart driving solution enables full-scenario autonomous navigation-assisted driving worldwide across diverse and fragmented environments such as urban roads, structured roads, and parking lots. It aims to provide customers with a comfortable, consistent, and efficient driving experience, addressing long-standing challenges in advanced autonomous driving applications, including poor generalization, inferior user experience, and high costs.
The A2000 multimodal AI large model employs a vision-based system and can also process multi-modal inputs such as camera and audio data, enabling the vehicle to see, hear, and understand its surroundings. This allows for diverse inference outputs including visual information, scene descriptions, and driving behaviors. Without relying on LiDAR or HD maps, the solution can provide advanced features such as city navigation, memory driving, highway navigation, and memory parking through pure-vision solutions combined with basic navigation information.
On December 26, 2024, Black Sesame Technologies launched the Huashan A2000 chip family, a high-performance chip platform designed for next-generation AI models. According to the company, the A2000 family integrates industry-leading multi-function units such as CPU (central processing unit), DSP (digital signal processor), GPU (graphics processing unit), NPU (neural processing unit), MCU (microcontroller unit), ISP (image signal processor), and CV (computer vision), enabling high integration and multi-task processing on a single chip. The chip also supports Transformer models.
Leveraging the A2000 family's abundant computing resources and versatile toolchain, Nullmax isdeploying its multimodal AI large model-enabled smart driving solution. The LMDAM (Large-Multimodal-Description-Action Model), which has been used in Nullmax Intelligence (the new-generation autonomous driving technology launched in July 2024), is set to be deployed on A2000 chip for real-car intallation.
Through the new in-depth collaboration, Nullmax and Black Sesame Technologies aim to build a next-generation intelligent driving solution. The combination of multimodal AI large model for smart drivng and the A2000 chip will provide cutting-edge autonomous driving applications and an exceptional experience, meeting the market demand for intelligent mobility.
Black Sesame Technologies, founded in 2016, is a leading provider of automotive-grade intelligent computing chips and chip-based solutions. The company has unveiled the Huashan series of high-performance chips, designed for autonomous driving, and the Wudang series for cross-domain computing, addressing the growing complexity of intelligent vehicle functions. The company's automotive-grade products and technologies empower smart vehicles with critical capabilities, including autonomous driving, smart cockpits, advanced imaging, and interconnectivity. Through in-house developed IP cores, algorithms, software-driven SoC, and SoC-based solutions, the company provides full-stack autonomous driving capabilities to meet the diverse needs of its customers.
Nullmax is an AI technology company specializing in autonomous driving applications. It is dedicated to creating full-scenario autonomous driving solutions and accelerating mobility transformation. Nullmax employs a machine learning-first technical approach and has developed a complete full-stack autonomous driving technology, the MAX. Based on this, it has constructed a cornerstone architecture comprising a data platform and an algorithm platform, providing a universal AI development platform that endows machines with human-like perception, learning, and interaction capabilities.
