ORB-based optimal lost robot self-recovery
Due to various reasons, an autonomous mobile robot may deviate from its planned trajectory. If the robot deviates beyond its knowledge of the environment, it is essentially ‘lost’. To complete its programmed task, it is thus important for the robot to be able to perform ‘lost-recovery’, which is to...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2017
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/71861 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Due to various reasons, an autonomous mobile robot may deviate from its planned trajectory. If the robot deviates beyond its knowledge of the environment, it is essentially ‘lost’. To complete its programmed task, it is thus important for the robot to be able to perform ‘lost-recovery’, which is to re-orientate itself back within the perimeters of a known environment. The key idea behind lost-recovery is the concept of simultaneous localization and mapping (SLAM), with place recognition being our area of interest. Existing place recognition techniques are based mostly on SIFT or SURF, which are highly accurate but computationally intensive.
This report will explore the integration of a recently-developed descriptor, ORB (Oriented Fast and Rotated Brief), and the DBoW2 hierarchical tree structure to create an algorithm which guides a lost robot back to its programmed path through visual means.
This report will start with a brief introduction about the lost robot problem, followed by a literature review on some of the existing work done on lost robot recovery and place recognition. Subsequent chapters will discuss in details the method proposed by the author and the results obtained from simulations in different environments. The report will conclude with analysis of results and suggestions for future work. |
|---|