SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights

By Michael Milford, Gordon Fraser Wyeth. IEEE International Conference on Robotics and Automation, 2012

1,015 citations recorded by the source corpus.

Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100% precision with recall rates of up to 60%.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/448babca3ceca4a4f2b2ce1d2e87cd4ff43e337a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Self-Driving Cars: A Survey (2019, methodology): “structed from sensor data during a mapping phase and stored for later use during the autonomous operation phase. Representations of the state space are often distinguished between topological [CUM08] [MIL12] [FORE18] and metric [HOR13] [MUT16] [SCH18] representations. Topological representations mod”
High-fidelity simulation for evaluating robotic vision performance (2016, methodology): “On the other hand, use of global image descriptors like BRIEFGIST [22] or patch-normalised downsampled images as in SeqSLAM [14] allows matching across change in conditions,”
Reliable Patch Trackers: Robust visual tracking by exploiting reliable patches (2015, background): “In model-free tracking, Grabner et al. [28] proposed a boosting classiﬁer to match the keypoint between different frames.”
Made to measure: Bespoke landmarks for 24-hour, all-weather localisation with a camera (2016, methodology): “SeqSLAM [1] and FAB-MAP [2] are examples of place recognition systems – they output the image (or place) to which the robot’s live image is most similar, however there is no information about the robot’s 6-DOF pose in the map.”
It's not easy seeing green: Lighting-resistant stereo Visual Teach & Repeat using color-constant images (2015, neutral): “Recent publications have provided the means for computing color-constant greyscale images from a three-channel camera [5], [14], and have been shown to be successful in localization and mapping systems [3], [8].”
Scan Context: Egocentric Spatial Descriptor for Place Recognition Within 3D Point Cloud Map (2018, methodology): “SeqSLAM [17] proposed the route-based approach and showed far improved performance than FAB-MAP.”
A review of multimodal image matching: Methods and applications (2021, methodology): “[24] proposed a method called SeqSLAM, which is used to visualize navigation under changing conditions.”
Patch-NetVLAD: Multi-Scale Fusion of Locally-Global Descriptors for Place Recognition (2021, background): “Place matches obtained through global descriptor matching are often re-ranked using sequential information [26, 85, 49], query expansion [31, 13], geometric verification [41, 27, 52] and feature fusion [83, 86].”
Robust visual SLAM across seasons (2015, background): “There has been a tremendous progress in the area of visual Simultaneous Localization and Mapping (SLAM) over the last couple of years, especially in the context of appearancebased place recognition, which is an essential building block to successfully find loop closures in the context of SLAM [6], [”
Are We There Yet? Challenging SeqSLAM on a 3000 km Journey Across All Four Seasons (2013, methodology): “Overall, the performance on the Nordland dataset is similar to the performance reported in [2] on two very different datasets.”
Convolutional Neural Network-based Place Recognition (2014, methodology): “SeqSLAM [Milford and Wyeth, 2012] directly uses pixel values to match image sequences and perform place recognition across extreme perceptual changes.”
Scan Context++: Structural Place Recognition Robust to Rotation and Lateral Variations in Urban Environments (2021, background): “Describing a place with a set of compact representations has been tackled in depth within the computer vision and robotics community, yielding many state-of-the-art visual place recognition meth-ods [4, 5, 6, 7].”
Intensity Scan Context: Coding Intensity and Geometry Relations for Loop Closure Detection (2020, background): “1) Temporal consistency check: In a SLAM system, it is observed that the occurrence of single loop closure often implies high similarity on the neighbour LiDAR scans since the sensor feedback is continuous in time [25].”
Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (2018, methodology): “We also evaluate the de-facto standard approach for loop-closure detection in robotics [23, 36], where robustness to changing conditions is critical for long-term autonomous navigation [17, 37, 46, 49, 66, 69]: FAB-MAP [20] is an image retrieval approach based on the Bag-ofWords (BoW) paradigm [62] ”