Perception & Localization · Full-time
VSLAM Algorithm Engineer
- Compensation
- Compensation negotiable
- Location
- Beijing
- Experience
- 3–5 years
- Education
- Master’s degree or above
Responsibilities
- Build multi-camera and IMU capture, hardware/software synchronization, timestamping, transport, buffering, and quality monitoring.
- Develop camera–IMU calibration workflows and evaluation for intrinsics, inter-camera and camera–IMU extrinsics, time offset, and rolling shutter.
- Improve VIO, VSLAM, localization, and trajectory recovery through feature tracking, IMU preintegration, nonlinear optimization, loop closure, and relocalization.
- Deploy on Ego devices and embedded Linux, evaluating compute, latency, memory, real-time behavior, and stability.
- Diagnose dropped frames, drift, calibration errors, exposure anomalies, IMU noise, and interrupted trajectories.
Requirements
- Master’s degree or above in computer vision, robotics, automation, surveying, electronic information, or AI.
- Proficient in C++ or Python and Linux, with strong algorithm and system engineering skills.
- Understand camera, fisheye/wide-angle, and IMU models, coordinate transforms, synchronization, and spatial calibration.
- Know VSLAM/VIO foundations: feature tracking, IMU preintegration, PnP, BA, nonlinear optimization, sliding windows, and loop closure.
- Hands-on sensor integration, data acquisition, debugging, and result analysis on real hardware.
Nice to have
- Experience with Jetson, ARM, Android, embedded Linux, GPU/NPU acceleration, or real-time debugging.
- Familiarity with ORB-SLAM3, VINS-Fusion, OpenVINS, OKVIS, Kimera, Kalibr, or OpenCalib.
- Experience with PTP, GMSL, MIPI, USB cameras, high-bandwidth multi-camera transport, or IMU drivers.
- Relevant publications at ICRA, IROS, RSS, CVPR, ICCV, ECCV, or CoRL.