AirLoop: Lifelong loop closure detection

Published: by
Dasong Gao

AirLoop: Lifelong loop closure detection

After deployment to the wild, the robot may benefit if it can incrementally learn from the working environments with newly observed data. However, as the perception algorithm involves more and more deep networks for improved robustness over traditional methods, this becomes hard: Deep models typically need to adjust all parameters jointly to adapt to new environments.

This poses a dilemma: On the one hand, if we only retrain on the data increments, fitting to the new data almost inevitably alters the parameters and, subsequently, the representations learned from the old data, leading to performance degradation over time. This phenomenon is referred to as catastrophic forgetting. On the other hand, if all previously observed data is retained to perform joint training after observing each new environment, the storage cost will grow linearly as the model learns from one environment to the next. This is prohibitive on resource-constrained devices such as a drone’s onboard computer.

A robot may encounter a series of new environments after deployment from which it can gradually lean a more suitable scene representation.

Lifelong loop closure detection

In this work, we focus on the specific problem of learning a loop closure detection (LCD) network in an incremental fashion. We developed AirLoop, a method that leverages techniques from lifelong learning to minimize forgetting when training loop closure detection models incrementally. To the best of our knowledge, AirLoop is one of the first work to study LCD in the lifelong learning context.

To achieve lifelong learning of LCD models, it is important to ensure that the geometry of the global descriptor space (i.e. descriptors’ pairwise similarities) does not deform as the network trains in more environments. For this purpose, we adopted the relational variant of two regularization losses for lifelong classification, dubbed RMAS (relational memory aware synapses) and RKD (relational knowledge distillation). The former operates by calculating the descriptor space’s “sensitivity” with respect to each parameter and use it to selectively penalize future changes to the parameters, whereas the later compare the output of the current network against its previous versions and protects the descriptor space from uncontrolled deformation.

Lifelong losses in AirLoop. RMAS and RKD regularizes the network in parameter space and embedding space, respectively.

Both method combined allows us to train the LCD model in a series of environments sequentially without suffering from serious catastrophic forgetting while only requiring constant memory and computation.


Here we show the lifelong learning outcome of the model on TartanAir, Nordland, and RobotCar:

Examples of loop closure detection on each dataset. Note that our model is able to handle cross-environment loop closure detection despite only trained in individual environments sequentially.

We also observe improved robustness after extended training:

Model performance comparison on TartanAir. The model is able to correct some of the previously made mistakes after training in more environments.



  • Dasong Gao, Chen Wang, and Sebastian Scherer. AirLoop: Lifelong Loop Closure Detection.” 2022 International Conference on Robotics and Automation (ICRA), 2022.

Please refer to our paper and code for details.


SLAM, Learning

Latest Research

AirIMU: Learning Uncertainty Propagation for Inertial Odometry
AirIMU: Learning Uncertainty Propagation for Inertial Odometry

AirIMU, a hybrid IO system designed for state estimation and uncertainty propagation.

SubT-MRS: Pushing SLAM Towards All-weather Environments
SubT-MRS: Pushing SLAM Towards All-weather Environments

Simultaneous localization and mapping (SLAM) is a fundamental task for numerous applications such...

TartanDrive 2.0: More Modalities and Better Infrastructure to Further Self-Supervised Learning Research in Off-Road Driving Tasks
TartanDrive 2.0: More Modalities and Better Infrastructure to Further Self-Supervised Learning Research in Off-Road Driving Tasks