June 2023
·
57 Reads
·
9 Citations
What is this page?
This page lists the scientific contributions of an author, who either does not have a ResearchGate profile, or has not yet added these contributions to their profile.
It was automatically created by ResearchGate to create a record of this author's body of work. We create such pages to advance our goal of creating and maintaining the most comprehensive scientific repository possible. In doing so, we process publicly available (personal) data relating to the author as a member of the scientific community.
If you're a ResearchGate member, you can follow this page to keep up with this author's work.
If you are this author, and you don't want us to display this page anymore, please let us know.
Publications (4)
June 2022
·
6 Reads
·
40 Citations
May 2022
·
4 Reads
·
11 Citations
September 2021
·
53 Reads
Loop closure detection is an important building block that ensures the accuracy and robustness of simultaneous localization and mapping (SLAM) systems. Due to their generalization ability, CNN-based approaches have received increasing attention. Although they normally benefit from training on datasets that are diverse and reflective of the environments, new environments often emerge after the model is deployed. It is therefore desirable to incorporate the data newly collected during operation for incremental learning. Nevertheless, simply finetuning the model on new data is infeasible since it may cause the model's performance on previously learned data to degrade over time, which is also known as the problem of catastrophic forgetting. In this paper, we present AirLoop, a method that leverages techniques from lifelong learning to minimize forgetting when training loop closure detection models incrementally. We experimentally demonstrate the effectiveness of AirLoop on TartanAir, Nordland, and RobotCar datasets. To the best of our knowledge, AirLoop is one of the first works to achieve lifelong learning of deep loop closure detectors.
Citations (3)
... The versatility of learning models has made them ubiquitous in robotics, now appearing in almost all layers of the modern software stack [67]. On the other hand, model-based optimization -the mainstay of traditional robotics -provides a level of robustness, accuracy and generalization that has proven difficult to match by learning-based methods [55]. ...
- Citing Conference Paper
June 2023
... Subsequently, these refined representations enable GNNs to perform various tasks such as predicting or classifying corresponding graphs, nodes, or edges. This mechanism facilitates the ability of GNN to better capture complex relationships and dependencies within graph-structured data, compared with traditional deep networks, which are limited to handling Euclidean data and thus may exhibit less precision in modeling intricate relationships within such data structures [6,14,52,69,77]. Furthermore, recent investigations have expanded the utilization of GNN to address multifaceted scenarios in graph data processing, including but not limited to large-scale, temporal, spatial, dynamic, and heterogeneous graphs, which has notably contributed to augmenting the efficacy and versatility of GNN [4,19,20,32,75,98,102,106]. ...
- Citing Conference Paper
June 2022
... Compared to mainstream methods at present, the system has improved short-term adaptation and long-term memory retention. [118] incorporates data from the new environment into incremental learning while minimizing amnesia during incremental training of the model by employing lifelong learning techniques. [119] proposed a maintenance method for an adaptive local map system that can intelligently eliminates redundant local maps to ensure the required robustness and stability of lifelong mapping and prevent the increase of global map rendering time and system memory usage caused by information redundancy. ...
- Citing Conference Paper
May 2022