感知AI算法工程师（佛山） Perception AI Algorithm Engineer

参与无人环卫车/环卫机器人感知 AI 算法研发，包括 2D/3D 目标检测、语 义/实例分割、多目标跟踪等，覆盖环卫典型目标。

参与多传感器融合感知（Camera+LiDAR 等）、BEV 感知、Occupancy 等 算法设计与实现，了解传感器标定流程，参与多传感器的内参、外参标定。

参与感知相关模型训练、评估、迭代、量化，在指导下实现 TensorRT/ONNX 等车端部署与性能优化。

参与数据闭环：参与并优化数据采集、标注规范、真值生成、难例挖掘、 指标体系建设，结合传感器标定结果优化数据质量，推动感知效果持续迭 代。

与定位、规控、硬件、测试团队协同，参与完成实车调试与量产落地，解 决传感器标定及感知相关实际工程问题

硕士及以上学历，计算机、电子、数学等相关专业。

具备比较扎实的深度学习与计算机视觉基础，掌握目标检测、分割、跟踪、 点云处理、多传感器融合中至少一个方向，熟悉 YOLO 等常规模型算法。

熟悉 C++/Python，掌握 OpenCV、PCL、ROS 等常用工具，能在指导下完 成多传感器标定方案设计、实施与优化，了解传感器标定与感知算法的联动优化 逻辑。

有实习或者参与量产项目落地经验优先：低速无人车、园区物流、环卫机器 人等均可。

了解感知部署与优化流程：模型量化、TensorRT、端侧推理性能调优。

具备良好的问题定位能力以及较强学习能力，善于沟通。 Job Responsibilities Take part in developing perception AI algorithms for autonomous sanitation vehicles and robots, including 2D/3D object detection, semantic/instance segmentation and multi-object tracking for typical sanitation targets. Design and implement multi-sensor fusion perception (Camera + LiDAR, etc.), BEV perception, Occupancy and other algorithms. Understand sensor calibration workflow and participate in intrinsic and extrinsic parameter calibration of multiple sensors. Conduct training, evaluation, iteration and quantization of perception models. Complete vehicle-side deployment and performance optimization based on TensorRT/ONNX under guidance. Engage in data closed-loop work: improve data collection, labeling standards, ground truth generation, hard case mining and indicator system construction. Optimize data quality with sensor calibration results to continuously upgrade perception performance. Cooperate with localization, planning & control, hardware and testing teams to carry out vehicle debugging and mass production implementation, and solve practical engineering issues related to sensor calibration and perception. Job Requirements Master’s degree or above, majoring in Computer Science, Electronics, Mathematics or related disciplines. Solid foundation in deep learning and computer vision. Have expertise in at least one field among object detection, segmentation, tracking, point cloud processing and multi-sensor fusion; familiar with mainstream models such as YOLO. Skilled in C++/Python. Experienced with common tools including OpenCV, PCL and ROS. Able to design, execute and optimize multi-sensor calibration schemes under guidance, and understand how sensor calibration interacts with perception algorithms. Prior internship or mass-production project experience (low-speed autonomous vehicles, campus logistics robots, sanitation robots, etc.) is preferred. Knowledgeable about perception deployment and optimization: model quantization, TensorRT, edge inference tuning. Strong problem diagnosis capability and quick learning ability, with good communication skills.