Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Scalable Graph-Based SLAM using Redundancy Detection
The goal of this project is to investigate approaches to quantify the information contributed by individual chunks of data in a SLAM graph, using information-theoretic measures, to boost the efficiency and scalability of SLAM.
Keywords: SLAM, Redundancy Detection, Information Theory, Mutual Information
State-of-the-art Simultaneous Localization And Mapping (SLAM) systems have reached a high level of maturity in single-agent [1] and multi-agent [2] scenarios. However, while showing substantial accuracy and robustness in various environments, the scalability of these solutions (in terms of participating agents and size of the workspace) remains a problem, since the amount of data that can be efficiently handled by the SLAM system is limited. One approach to mitigate this problem is to employ redundancy detection scheme to sparsify the SLAM graph, since not every chunk of data in the graph contributes the same amount of information (such as data from repeated visits of the same location). As shown in [3], redundancy detection based on information-theoretic notions is able to significantly compress SLAM graphs with minimal loss in accuracy, furthermore outperforming hand-crafted heuristics. However, these approaches are currently still too time-consuming to run online during the SLAM session. The goal of this project is, based on the results in [3], to investigate more efficient and accurate approaches for redundancy detection and removal on SLAM graphs, in order to further boost the applicability of these approaches and the scalability of contemporary SLAM solutions.
References:
[1] Qin et al.: VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator, T-RO 2018
[2] Karrer et al.: CVI-SLAM – Collaborative Visual-Inertial SLAM, RA-L 2018
[3] Schmuck and Chli: On the redundancy detection in keyframe-based SLAM, 3DV 2019
[4] Mazuran et al.: Nonlinear factor recovery for long-term SLAM, IJRR 2016
State-of-the-art Simultaneous Localization And Mapping (SLAM) systems have reached a high level of maturity in single-agent [1] and multi-agent [2] scenarios. However, while showing substantial accuracy and robustness in various environments, the scalability of these solutions (in terms of participating agents and size of the workspace) remains a problem, since the amount of data that can be efficiently handled by the SLAM system is limited. One approach to mitigate this problem is to employ redundancy detection scheme to sparsify the SLAM graph, since not every chunk of data in the graph contributes the same amount of information (such as data from repeated visits of the same location). As shown in [3], redundancy detection based on information-theoretic notions is able to significantly compress SLAM graphs with minimal loss in accuracy, furthermore outperforming hand-crafted heuristics. However, these approaches are currently still too time-consuming to run online during the SLAM session. The goal of this project is, based on the results in [3], to investigate more efficient and accurate approaches for redundancy detection and removal on SLAM graphs, in order to further boost the applicability of these approaches and the scalability of contemporary SLAM solutions.
References:
[1] Qin et al.: VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator, T-RO 2018
[3] Schmuck and Chli: On the redundancy detection in keyframe-based SLAM, 3DV 2019
[4] Mazuran et al.: Nonlinear factor recovery for long-term SLAM, IJRR 2016
Not specified
- Excellent mathematical skills (especially in non-linear optimization, probability and information theory)
- High degree of autonomy, high self-motivation, research-driven attitude
- Programming experience (ideally C++)
- Excellent mathematical skills (especially in non-linear optimization, probability and information theory) - High degree of autonomy, high self-motivation, research-driven attitude - Programming experience (ideally C++)