The rapid development of autonomous driving technology is leading us into a new era of intelligence. At the recent Beijing Zhiyuan Conference, Zhang Yaqin, academician of the Chinese Academy of Engineering and dean of the Intelligent Industry Research Institute (AIR) of Tsinghua University, a senior expert in the field of unmanned driving, put forward a bold prediction: "2025 will become the first year of embodied intelligence, and unmanned driving will surpass human drivers." This prediction not only points out the direction for the development of autonomous driving technology, but also brings new expectations to the entire field of artificial intelligence.
The rise of embodied intelligence
Embodied Artificial Intelligence (EAI), as an emerging branch of artificial intelligence, emphasizes the interaction between agents and the physical world. This agent is not only able to perceive the environment, but also interact with the environment through self-learning to perform complex tasks. Driverless cars are a typical example of the concept of embodied intelligence, which integrates advanced sensors, computer vision, machine learning and other technologies to achieve perception, decision-making and response to complex traffic environments.
Pictured: Zhang Yaqin, academician of the Chinese Academy of Engineering and dean of AIR at Tsinghua University, predicted: "2025 will become the first year of embodied intelligence, and unmanned driving will surpass human drivers." (Picture from the Internet)
The current situation and progress of autonomous driving technology
Driverless technology has made significant progress in recent years. Academician Zhang Yaqin pointed out that the current safety of unmanned driving has been at least 10 times higher than that of human driving. This achievement is due to the validation of a large number of real vehicle tests and simulation environments, as well as the continuous optimization of big data analysis and machine learning algorithms. For example, driverless cars from companies such as Waymo and Baidu Apollo have been approved for commercial operation in select cities, marking the beginning of the transition from the experimental stage to commercial application of driverless technology.
The role of large-scale model technology in promoting unmanned driving
The application of large model technology in the field of unmanned driving provides a new idea for solving the long-tail problem and data generation problem. Baidu has amassed 100 million kilometers of driving data and generated a large amount of data for training through generative AI technology. These data not only improve the intelligence level of the unmanned driving system, but also provide strong support for the application of unmanned driving in actual scenarios.
Challenges of autonomous driving
While autonomous technology has made significant progress, there are still some challenges. Academician Zhang Yaqin pointed out that driverless cars may be too disciplined in the driving process, lacking the flexibility and experience of human drivers. For example, driverless cars may not overtake or change lanes at the right time, which can affect driving efficiency in complex traffic environments. In addition, driverless cars may also have limitations when it comes to handling some extreme or rare traffic situations.
The future of autonomous driving
Academician Zhang Yaqin's prediction depicts a future picture for us of unmanned driving technology that surpasses human drivers. By 2025, as the technology continues to mature and regulations gradually improve, driverless cars are expected to be commercialized in more cities and regions. At the same time, the application of unmanned driving technology will also be expanded to more fields, such as logistics and distribution, public transportation, intelligent travel, etc.
To sum up, autonomous driving technology, as a representative of embodied intelligence, is gradually changing our travel methods and transportation systems. With the continuous advancement of technology and the continuous expansion of application scenarios, autonomous driving is expected to achieve a wider range of commercial applications in the next few years. At the same time, the industry also needs to continue to explore and solve the technical and social challenges faced by autonomous driving to promote the comprehensive development of embodied intelligence. 2025 may indeed become the first year of embodied intelligence, let's wait and see.