IMPLEMENTATION OF SIMULTANEOUS LOCALIZATION AND MAPPING ON AUTONOMOUS VEHICLE USING ORB-SLAM2
Autonomous vehicle is an alternative solution to reduce the number of accidents due to human-error. Autonomous vehicle conduct perception to know its dynamic state such as position and velocity. Unaccurate localization of autonomous vehicle can cause wrong decision and accident. Global Navigation Sa...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Subjects: | |
| Online Access: | https://digilib.itb.ac.id/gdl/view/55314 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Autonomous vehicle is an alternative solution to reduce the number of accidents due to human-error. Autonomous vehicle conduct perception to know its dynamic state such as position and velocity. Unaccurate localization of autonomous vehicle can cause wrong decision and accident. Global Navigation Satellite System (GNSS) is often used as localization sensor but GNSS measurement give unaccurate localization in close area such as under building or trees. In the other hand, autonomous vehicle navigate in various area including close area. Thus, it is necessary to develop localization method that is give accurate estimation in various area. This research is part of autonomous vehicle project in Institut Teknologi Bandung. The project teams has designed an autonomous vehicle but its localization only depend GNSS which is its accuration are affected by environment condition.
This research will implement and develop Simultaneous Localization and Mapping (SLAM) using ORB (Oriented FAST and Rotated BRIEF) method on autonomous vehicle to give more accurate localization in various area. Simultaneous Localization and Mapping will use one of visual-SLAM method that is ORB with RGB-D camera as sensor. Visual-SLAM estimate position using place recognition and depth estimation on image that is received by RGB-D camera. Thus visual-SLAM can give relative accurate localization in close area.
ORB-SLAM2 is relatively faster on computing and rebust of sunlight than the others visual-SLAM method. ORB-SLAM2 then will be implemented and developed on golf cart which have been modified as autonomous vehicle and RGB-D camera as sensor. This research will discuss evaluation and necessary modification to implement ORB-SLAM2 on autonomous vehicle and using RGB-D camera as sensor. |
|---|