智驾系统架构师_XC

Company Description

Do you want beneficial technologies being shaped by your ideas? Whether in the areas of mobility solutions, consumer goods, industrial technology or energy and building technology - with us, you will have the chance to improve quality of life all across the globe. Welcome to Bosch.

Job Description

1. System Architecture Design: Develop the algorithm system architecture for urban mapless E2E platforms, leading the design of the full-link software architecture from raw sensor data input to vehicle control command output. Define clear module boundaries, module functions and tasks, and interface protocols and data flows between algorithm modules. Review and document design changes using UML.

2. Algorithm Solution Review: Review technical solutions for core algorithms such as perception, mapping, prediction, and planning, evaluating engineering feasibility and compatibility with existing architectures, and proposing improvements for algorithm modules.

3. Performance Evaluation Establishment: Establish a multi-category scenario set for evaluating system functionality and performance. Based on the module function link architecture, assess performance improvements and regressions between different releases, and propose quantitative modification requirements and architectural change demands for modules.

4. Cross-team Collaboration: Strong driving and coordination abilities, capable of close collaboration with TPMs, development teams for various modules (perception, prediction, planning, control, underlying systems), and testing teams to unify technical understanding and ensure successful implementation of architectural solutions.

Qualifications

1. Bachelor's degree or above in Electronic Information, Vehicle Engineering, Computer Science, Software Engineering, etc.

2. Programming Languages: Proficient in C++, with over 3 years of experience in embedded complex system development. Familiar with ROS/ROS2 frameworks and proficient in using Linux development systems.

3. At least one complete participation in the engineering implementation of an urban mapless NOA project, leading the module algorithm architecture design within it.

4. Toolchain: Proficient in Docker, Git. Familiar with simulation platforms (e.g., CARLA, LGSVL), data platforms, and model training frameworks (e.g., PyTorch, TensorFlow).

5. At least one of the following module development processes has been led:

   1. Familiar with the application of deep learning models (e.g., CNN, RNN, Transformer) in computer vision, including object detection (e.g., YOLO, Faster R-CNN, DETR), semantic segmentation (e.g., SegNet, DeepLab), instance segmentation, lane detection, and drivable area segmentation.
   2. Familiar with BEV (Bird's Eye View) perception technology, such as BEVFormer, LSS (Lift, Splat, Shoot), and temporal fusion technology.
   3. Familiar with environmental modeling techniques, including Grid Map, Occupancy Network, Dynamic Scene Graph, etc.
   4. Familiar with common motion planning algorithms such as A*, D*, RRT*, Lattice Planner, EM Planner, and optimization-based planning methods (e.g., MPC).
   5. Familiar with the application of reinforcement learning (RL) or imitation learning (IL) in two-stage planning, as well as problems and cutting-edge advancements in end-to-end planning engineering applications.