⾃动驾驶算法研究与开发（战略实习⽣）

基于项目需求，调研自动驾驶行为预测、轨迹生成、潜空间建模、生成式模型、运动规划以及安全决策等相关技术路线。 探索基于 Latent Model 的自动驾驶建模方法，包括语义潜空间建模、行为意图表示、风险感知行为编码、可控轨迹生成以及先验 - 后验表示对齐等方向。 探索生成式轨迹模型与 Planner 的结合方式，包括候选轨迹生成、轨迹打分、风险评估、安全约束建模以及规划结果优化等。 探索行为生成与规划算法在安全关键驾驶功能中的应用，如 AEB、AES、碰撞规避、风险感知机动生成以及紧急避险轨迹规划等场景。 完成相关算法的实现、训练与评估，在自动驾驶数据集或仿真平台上开展实验，包括基线复现、消融实验、指标评估、结果可视化以及失败案例分析。 汇总研究和开发成果，形成专利、技术报告以及学术论文。 及时反馈开发进度和遇到的问题，与团队积极讨论解决方案，并记录关键实验结果、技术结论和经验沉淀

具备以下任一方向的研究或项目经历：深度学习、计算机视觉、自动驾驶、轨迹预测、运动规划、生成式模型或 Latent Model 相关方法。 具备较扎实的机器学习与深度学习基础。 具备良好的编程思维和工程实现能力，熟悉 Python、PyTorch 以及 AI-assisted coding / Vibe Coding 工作流者优先。 熟悉至少一种相关工具、数据集或平台者优先，如 OpenCV、ROS、CARLA、nuScenes、Waymo、Argoverse、nuPlan 或类似资源。 计算机、人工智能、自动化、机器人、电子工程、数学、车辆工程或相关理工科专业背景。 乐于学习新的研究方向、工具和平台，具备较强的独立性、主动性和问题解决能力。 具备良好的英文读写能力，尤其是学术论文阅读和技术文档整理能力。 能尽快入职者优先，能够连续实习 6 个月以上者优先。 Responsibilities: Investigate feasible technologies based on project requirements, including behavior prediction, trajectory generation, latent representation learning, generative models, motion planning, and safety-aware decision-making for autonomous driving. Explore latent model-based methods for autonomous driving, including semantic latent space modeling, intent representation, risk-aware behavior encoding, controllable trajectory generation, and prior-posterior representation alignment. Explore the integration of generative trajectory models with planning modules, including candidate trajectory generation, trajectory scoring, risk evaluation, safety constraint modeling, and planner output refinement. Investigate the application of behavior generation and planning algorithms in safety-critical driving functions, such as AEB, AES, collision avoidance, risk-aware maneuver generation, and emergency trajectory planning. Implement, train, and evaluate relevant algorithms on autonomous driving datasets or simulation platforms. Conduct experiments including baseline reproduction, ablation studies, metric evaluation, visualization, and failure case analysis. Synthesize research and development outcomes into patents, technical reports, and academic papers. Provide timely feedback on development progress and technical issues, actively discuss solutions with the team, and document key findings, experiments, and conclusions. Requirements: Experience with at least one of the following areas: deep learning, computer vision, autonomous driving, trajectory prediction, motion planning, generative models, or latent model-based methods. Solid understanding of machine learning and deep learning fundamentals. Possess strong programming thinking and engineering implementation ability. Familiarity with Python, PyTorch, and AI-assisted coding / Vibe Coding workflows is preferred. Familiarity with at least one related tool, dataset, or platform, such as OpenCV, ROS, CARLA, nuScenes, Waymo, Argoverse, nuPlan, or similar resources, is preferred. Major in computer science, artificial intelligence, automation, robotics, electronic engineering, mathematics, vehicle engineering, or related fields. Willing to learn new research directions, tools, and platforms. Self-dependent, highly motivated, and with a problem-solving mentality. Good English reading and writing capability is required, especially for academic paper reading and technical documentation. Ability to join as soon as possible is preferred. Candidates who can stay for more than 6 months will be strongly preferred.