is SLAM a doomed approach in agri fields?

[jediofgever]

In the last decade, the interest in using fully autonomous mobile robots for agricultural tasks has been growing significantly. Agricultural environments are highly visual repetitive and present high dynamic scenes because of the movement of the leafs of the field caused by the wind. These features, among others, make the agricultural environment a very strong challenge for vision-based SLAM systems. In this work, we assess the well-known S-PTAM and ORB-SLAM2 Visual SLAM systems and the Visual-Inertial SLAM S-MSCKF in agricultural environments. In particular the evaluation is performed on the recently released Rosario dataset. The evaluation shows that the three systems achieve a poor performance in terms of accuracy and robustness in contrast to the performance reported on urban or indoor environments where they are usually tested.

https://www.researchgate.net/publication/343402703_Evaluation_of_Visual_SLAM_Algorithms_on_Agricultural_Dataset

[jediofgever]

This paper demonstrates a system capable of combining a sparse, indirect, monocular visual SLAM, with both offline and real-time Multi-View Stereo (MVS) reconstruction algorithms. This combination overcomes many obstacles encountered by autonomous vehicles or robots employed in agricultural environments, such as overly repetitive patterns, need for very detailed reconstructions, and abrupt movements caused by uneven roads. Furthermore, the use of a monocular SLAM makes our system much easier to integrate with an existing device, as we do not rely on a LiDAR (which is expensive and power consuming), or stereo camera (whose calibration is sensitive to external perturbation e.g. camera being displaced). To the best of our knowledge, this paper presents the first evaluation results for monocular SLAM, and our work further explores unsupervised depth estimation on this specific application scenario by simulating RGB-D SLAM to tackle the scale ambiguity, and shows our approach produces reconstructions that are helpful to various agricultural tasks. Moreover, we highlight that our experiments provide meaningful insight to improve monocular SLAM systems under agricultural settings.

https://arxiv.org/pdf/2011.01122.pdf

[jediofgever]

There are places where SLAM could prove to be useful. But it seems that the nature of agri fields would require SLAM to be supported by additional setups.

GPS-SLAM_An_Augmentation_of_the_ORB-SLAM_Algorithm

The paper replaces some parts of the algo with GPS-IMU. The modifications can be summarized as ;

• ORB-SLAM2 uses TrackWithMotionModel in order to predict next pose of camera, it bases on transformation matrices between last two poses of frames. This works well if movement of camera was not large. In case of low frame rate does not work so well. The paper suggests to replace TrackWithMotionModel with TrackWithGPSData which is more consistent with low frame rate.

• The paper suggests another modification on TrackReferenceKeyFrame . This part is interesting because it tries to deal with aggressive turns, these are the cases where ORB-SLAM 2 suffers.

[github-actions[bot]]

Stale issue message