Development of robust 3D SLAM & object detection software for the harsh environment

Date
Mar. 2020 - Feb. 2022

Project Objective

Develop real-time position estimation and 3D map generation algorithm for unmanned vehicles traveling in the wilderness environment based on lidar, camera, and IMU sensor fusion data
Develop an algorithm that generates location information along with object recognition and classification information by learning from existing datasets in various environments.
Developed algorithms to convert camera-based recognition/classification information into 3D location information based on calibration information of camera and lidar sensors
Verified performance through interworking with previously developed unmanned vehicles

Contributions

Investigated, compared, selected, analyzed, and implemented lidar-based location estimation algorithm
Investigated, compared, selected, and analyzed camera-based position estimation algorithms
Implemented multi-sensor fusion module
Factor graph-based sensor fusion → To improve the quality of mapping results
Kalman filter-based sensor fusion → To improve the speed of Localization results
Implement obstacle occupancy grid mapping module
Analyzed the results and created a user manual

Thesis

[IPIU2021] 야지 환경에 강인한 다중 주행 거리계 융합 기반 3차원 동시적 위치 추정 및 지도 작성.pdf
3723.9KB