I Can See for Miles and Miles: An Extended Field Test of Visual Teach and Repeat 2.0

Paton, Michael; MacTavish, Kirk; Berczi, Laszlo-Peter; Es, Sebastian Kai van; Barfoot, Timothy D.

doi:10.1007/978-3-319-67361-5_27

Cited by 24 publications

(19 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When evaluating feature matching, we assumed that we had a prior on the vehicle's topological location in the map, such that we could reliably identify the nearest vertex in the pose graph. This is typical for autonomous visual route-following systems such as [4], where the topological prior is derived by dead reckoning from a previous successful localization, or using place recognition or GNSS in the event that the system becomes lost.…”

Section: Methodsmentioning

confidence: 99%

“…1 Space and Terrestrial Autonomous Robotic Systems (STARS) Lab, University of Toronto Institute for Aerospace Studies (UTIAS), Canada. 2 Long-term maps based on multiple visual 'experiences' of an environment have proven to be effective tools for metric localization through daily and seasonal appearance change [1]- [4]. In [3], consecutive visual experiences are recorded in a spatio-temporal pose graph, and localization against a privileged mapping experience proceeds by recalling a relevant experience and tracing through a chain of relative transformations in the graph.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Learning Matchable Image Transformations for Long-Term Metric Visual Localization

Clement

Gridseth²,

Tomasi

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Long-term metric self-localization is an essential capability of autonomous mobile robots, but remains challenging for vision-based systems due to appearance changes caused by lighting, weather, or seasonal variations. While experience-based mapping has proven to be an effective technique for bridging the 'appearance gap,' the number of experiences required for reliable metric localization over days or months can be very large, and methods for reducing the necessary number of experiences are needed for this approach to scale. Taking inspiration from color constancy theory, we learn a nonlinear RGB-to-grayscale mapping that explicitly maximizes the number of inlier feature matches for images captured under different lighting and weather conditions, and use it as a pre-processing step in a conventional single-experience localization pipeline to improve its robustness to appearance change. We train this mapping by approximating the target non-differentiable localization pipeline with a deep neural network, and find that incorporating a learned low-dimensional context feature can further improve cross-appearance feature matching. Using synthetic and realworld datasets, we demonstrate substantial improvements in localization performance across day-night cycles, enabling continuous metric localization over a 30-hour period using a single mapping experience, and allowing experience-based localization to scale to long deployments with dramatically reduced data requirements.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Learning Matchable Image Transformations for Long-Term Metric Visual Localization

Clement

Gridseth²,

Tomasi

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

show abstract

“…The Ethier field test presented by Paton, MacTavish, et al () demonstrated the VT&R 2.0 system’s ability to navigate large‐scale networks across intraseasonal appearance change. However, the size of the Ethier network meant that the amount of experiences gathered on any path was relatively small.…”

Section: Methodsmentioning

confidence: 99%

“…This paper builds upon our earlier work which introduces MEL for VT&R 2.0 with results on an offline data set (Paton, MacTavish, Warren, & Barfoot, ), and an initial field test with over 140 autonomous km on a 5‐km network of paths (Paton, MacTavish, Berczi, van Es, & Barfoot, ). This system stores new experiences in a spatiotemporal pose graph (STPG), with local–metric accuracy and global topology.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Selective memory: Recalling relevant experience for long‐term visual localization

MacTavish

Paton

Barfoot

2018

Journal of Field Robotics

Self Cite

View full text Add to dashboard Cite

Visual navigation is a key enabling technology for autonomous mobile vehicles. The ability to provide large-scale, long-term navigation using low-cost, low-power vision sensors is appealing for industrial applications. A crucial requirement for long-term navigation systems is the ability to localize in environments whose appearance is constantly changing over time-due to lighting, weather, seasons, and physical changes. This paper presents a multiexperience localization (MEL) system that uses a powerful map representation-storing every visual experience in layers-that does not make assumptions about underlying appearance modalities and generators. Our localization system provides real-time performance by selecting online, a subset of experiences against which to localize. We achieve this task through a novel experience-triage algorithm based on collaborative filtering, which selects experiences relevant to the live view, outperforming competing techniques. Based on classical memory-based recommender systems, this technique also enables landmarklevel recommendations, is entirely online, and requires no training data. We demonstrate the capabilities of the MEL system in the context of long-term autonomous path following in unstructured outdoor environments with a challenging 100-day field experiment through day, night, snow, spring, and summer. We furthermore provide offline analysis comparing our system to several state-of-theart alternatives. We show that the combination of the novel methods presented in this paper enable full use of incredibly rich multiexperience maps, opening the door to robust long-term visual localization.

show abstract

Adaptive Image Processing Methods for Outdoor Autonomous Vehicles

Halodová

Dvořáková

Majer

et al. 2019

Modelling and Simulation for Autonomous Systems

View full text Add to dashboard Cite

I Can See for Miles and Miles: An Extended Field Test of Visual Teach and Repeat 2.0

Cited by 24 publications

References 13 publications

Learning Matchable Image Transformations for Long-Term Metric Visual Localization

Learning Matchable Image Transformations for Long-Term Metric Visual Localization

Selective memory: Recalling relevant experience for long‐term visual localization

Adaptive Image Processing Methods for Outdoor Autonomous Vehicles

Contact Info

Product

Resources

About