人工智能的发展如火如荼，自动驾驶汽车作为人工智能在汽车产业和交通领域的深层次应用，是汽车产业迭代升级的新发展方向。但是自动驾驶汽车在降低人们驾驶负担、提升驾乘体验、满足市场需求的同时，不可避免地给交通出行带来了新的风险。目前，国内外已经出现了不少自动驾驶汽车交通肇事的案例，引发了学界关于其刑事归责的新议题。不同于传统汽车，对自动驾驶汽车交通肇事进行刑事归责存在着责任主体资格、注意义务范围的界定等问题，导致归责法理、归责逻辑发生变化，为执法与司法带来了诸多挑战。因此，需要明确自动驾驶模式下汽车交通肇事的刑事归责之标准，实现自动驾驶模式下汽车交通肇事与刑事法律的有效衔接，以回应行业发展的需求。

本文主要通过文献研究法、跨学科研究法、规范分析法，从四个方面展开分析：

首先，以自动驾驶发展现状的考察入手，从自动驾驶相关概念出发，明确界定本文研究对象，梳理国内外自动驾驶的发展脉络及立法现状。

其次，从理论与实践两个方面分析自动驾驶汽车交通肇事对刑事归责带来的挑战。其中，理论争议主要集中在责任主体资格的认定、注意义务范围的界定以及因果关系的判断上；实践层面表现在由于缺乏统一的裁判路径，对自动驾驶汽车交通肇事案件的刑事认定没有形成统一的“度量衡”，基于不同的情况和学说，法官会做出不同的判决。

第三，通过否定自动驾驶汽车刑事责任主体资格与肯定人类驾驶员和自动驾驶汽车生产商刑事责任主体资格、借鉴信赖原则界定注意义务范围、引入客观归责论明确因果关系判断规则三个层面，对自动驾驶交通肇事的刑事归责标准进行深入分析，提出建议。

最后，将自动驾驶模式分为辅助驾驶、人机共驾、无人驾驶三个阶段，对前述标准进行具体应用：在Lv0-Lv2辅助驾驶阶段下，行驶过程与传统驾驶几乎不存在差别，按照对传统驾驶交通肇事的规制办法处理；在Lv3人机共驾阶段下，当驾驶员主导驾驶时，比照Lv0-Lv2辅助驾驶阶段处理，当自动驾驶汽车主导驾驶时，驾驶员关于及时接管注意义务的履行情况便成为了驾驶员与生产商之间责任分配的关键所在；在Lv4-Lv5无人驾驶阶段，自动驾驶汽车的实际控制者是自动驾驶系统本身，驾驶员无注意义务，一般由自动驾驶汽车生产商对交通肇事的危害后果承担过失责任。

本文希望通过对刑事归责标准的研究，厘清不同等级自动驾驶模式下汽车交通肇事的刑事责任主体及责任分配，为自动驾驶模式下汽车交通肇事进行刑事归责提供新的解决路径。

The development of artificial intelligence is in full swing. As a deep application of artificial intelligence in the automobile industry and the field of transportation, autonomous vehicle is a new development direction of the iterative upgrading of the automobile industry. While reducing people's driving burden, improving driving experience, and meeting market demand, autonomous vehicles inevitably bring new risks to transportation. At present, there have been many cases of traffic accidents caused by autonomous vehicle at home and abroad, which have triggered new issues in the academic circle regarding their criminal attribution. Different from traditional cars, there are some problems in the criminal imputation of autonomous vehicle for traffic accidents, such as the qualification of the responsible subject and the definition of the scope of duty of care, which lead to changes in the imputation jurisprudence and imputation logic, and bring many challenges to law enforcement and justice. Therefore, it is necessary to clarify the standards for criminal imputation of autonomous vehicle traffic accidents, and realize the effective connection between automobile traffic accidents and criminal laws in autonomous driving mode to respond to the needs of industry development.

Through literature research method, interdisciplinary research method and normative analysis method, this paper analyzes this research from the following three aspects:

Firstly, starting from the investigation of the development status of autonomous driving, starting from the concept of autonomous driving, the research object of this paper is clearly defined, and the development context and legislative status of autonomous driving at home and abroad are sorted out.

Secondly, this paper analyzes the challenges brought by autonomous vehicle traffic accidents to criminal liability from both theoretical and practical aspects. Among them, the theoretical disputes mainly focus on the identification of the qualification of the responsible subject, the definition of the scope of duty of care and the judgment of causality. In practice, due to the lack of unified judgment standards, there is no unified "metric" for the criminal determination of autonomous vehicle traffic accidents, and judges will make different judgments based on different situations and doctrines.

Thirdly, by denying the qualification of autonomous vehicle criminal liability subjects and affirming the criminal liability subject qualification of autonomous vehicle manufacturers, drawing on the principle of reliance to define the scope of the duty of care, and introducing the theory of objection imputation to clarify the rules of causation, this paper provides an in-depth analysis of the criminal imputation standards for autonomous vehicle traffic accidents and makes suggestions.

Finally, this paper applies the above standard by dividing Lv0-Lv5 autonomous driving mode into three stages: Lv0-Lv2 assisted driving, Lv3 "man-machine co-driving" and Lv4-Lv5 unmanned driving. In the Lv0-Lv2 assisted driving stage, there is almost no substantial difference between the driving process and traditional driving, so it can be handled in accordance with the regulation of traditional driving traffic accidents. In the Lv3 "man-machine co-driving" stage, when the driver leads the driving, it is handled according to the Lv0-Lv2 assisted driving stage, and when the autonomous vehicle dominates the driving, the driver's performance of the duty of care in time becomes the key to the distribution of responsibilities between the driver and the manufacturer. In the Lv4-Lv5 autonomous driving stage, the actual controller of the autonomous vehicle is the automatic driving system itself, the driver has no duty of care, and the autonomous vehicle manufacturer is generally liable for the harmful consequences of the traffic accident.

This paper hopes to provide a new solution for criminal imputation of automobile traffic accidents in autonomous driving mode by studying the standards and clarifying the criminal liability subjects and liability allocation of automobile traffic accidents in different levels of autonomous driving mode.