- Deliver features on an automatic tool for the High-Definition(HD) map data
collection. The tool helps this team gather city-scale mapping data (approximately
3000 kilometers) by a part-time driver. The features include the sensors'
disconnection handler and an interactive Leaflet map viewer showing the instant
location and past trajectories.
- Researched the global accuracy criteria of HD maps. Implemented constraints using
pose-graph optimization to elevate the absolute accuracy of the HD maps.
- Implement a pose-graph optimization library using Ceres-Solver to accelerate the
processes before HD maps export. The optimization helps the lidar scans match and
connect pieces of scan points into a HD maps.
- Manage, generate, and annotate the constraints used in HD maps for elevating the
scan-matching quality. These maps are vital in the team's controller module and
serve many purposes, like traffic light detection. I have participated in 200+ km of
map production in Taiwan. Moreover, I have contributed to 130+ km on the city link
in Melbourne, Australia.
Project
This section is still constructed...
More projects will be here soon.
[WACV 2026] SceneEdited: A City-Scale Benchmark for 3D HD
Map Updating via Image-Guided Change Detection
keywords: Dataset, HD Map maintenance, Change Detection