城区静态感知算法专家_XC
Core
Develop road structure cognition and mapless static perception algorithms for urban NOA, reconstructing road topology, lane lines, curbs, and road markings from vehicle vision to support planning decisions without reliance on high-definition maps.
Role type
Senior IC autonomous driving perception algorithm engineer (urban static perception)
Builds
Production-ready perception models and vectorized map representations for urban driving scenarios
Domain
Autonomous driving / Computer Vision / Urban Navigation
Deliverable
production ML models
Required skills
Deep learning (Transformer, CNN), BEV perception, Lane topology inference, Multi-task learning, End-to-end perception, Python, C++, Model deployment on embedded platforms
Preferred skills
Topology frameworks (Topo2Seq, TopoHR, TopoStreamer), SLAM, Multi-view geometry, Non-linear optimization, End-to-end driving architecture, Model quantization
Technologies
BEVFormer, MapTR, TopoNet, LaneSegNet, Occupancy Network, World Model, VLA,征程6M
Responsibilities
Develop lane line detection and modeling under complex conditions; Develop curb detection and geometric parameterization; Detect and recognize road markings (stop lines, arrows, etc.); Identify and understand intersection types and connectivity; Build lane topology from multi-view images without HD maps; Fuse navigation and crowdsourced map priors for topology inference; Develop long-term temporal models for stability; Generate end-to-end driving guidance lines; Optimize and deploy unified perception models on car-grade embedded chips; Track and apply frontier technologies (BEV, Occupancy, VLA) to production projects
Seniority
Senior, hands-on IC