no code implementations • 15 May 2023 • Jonas Frey, Matias Mattamala, Nived Chebrolu, Cesar Cadena, Maurice Fallon, Marco Hutter
We demonstrate the advantages of our approach with experiments and ablation studies in challenging environments in forests, parks, and grasslands.
no code implementations • CVPR 2023 • Zhizheng Liu, Francesco Milano, Jonas Frey, Roland Siegwart, Hermann Blum, Cesar Cadena
Due to the mismatch between training and deployment data, adapting the model on the new scenes is often crucial to obtain good performance.
1 code implementation • 16 Sep 2022 • Samuel Looper, Javier Rodriguez-Puigvert, Roland Siegwart, Cesar Cadena, Lukas Schmid
In this work, we formalize the task of semantic scene variability estimation and identify three main varieties of semantic scene change: changes in the position of an object, its semantic state, or the composition of a scene as a whole.
1 code implementation • 17 Aug 2022 • Lukas Schmid, Mansoor Nasir Cheema, Victor Reijgwart, Roland Siegwart, Federico Tombari, Cesar Cadena
We further present an informative path planning method, leveraging the capabilities of our mapping approach and a novel scene-completion-aware information gain.
1 code implementation • 21 Jun 2022 • Hermann Blum, Marcus G. Müller, Abel Gawel, Roland Siegwart, Cesar Cadena
In order to operate in human environments, a robot's semantic perception has to overcome open-world challenges such as novel objects and domain gaps.
no code implementations • 3 May 2022 • Michael Pantic, Cesar Cadena, Roland Siegwart, Lionel Ott
This work investigates the use of Neural implicit representations, specifically Neural Radiance Fields (NeRF), for geometrical queries and motion planning.
no code implementations • 1 Mar 2022 • Chunwei Xing, Xinyu Sun, Andrei Cramariuc, Samuel Gull, Jen Jen Chung, Cesar Cadena, Roland Siegwart, Florian Tschopp
However, handcrafted topological descriptors are hard to tune and not robust to environmental noise, drastic perspective changes, object occlusion or misdetections.
1 code implementation • 18 Oct 2021 • Stefan Lionar, Lukas Schmid, Cesar Cadena, Roland Siegwart, Andrei Cramariuc
We present a novel 3D mapping method leveraging the recent progress in neural implicit representation for 3D reconstruction.
1 code implementation • 6 Oct 2021 • Boyang Sun, Jiaxu Xing, Hermann Blum, Roland Siegwart, Cesar Cadena
The proposed framework infers task failures by evaluating the individual prediction, across multiple visual perception tasks for different regions in an image.
no code implementations • 20 Sep 2021 • Florian Tschopp, Juan Nieto, Roland Siegwart, Cesar Cadena
Introducing semantically meaningful objects to visual Simultaneous Localization And Mapping (SLAM) has the potential to improve both the accuracy and reliability of pose estimates, especially in challenging scenarios with significant view-point and appearance changes.
1 code implementation • 4 May 2021 • Hermann Blum, Francesco Milano, René Zurbrügg, Roland Siegward, Cesar Cadena, Abel Gawel
We find memory replay an effective measure to reduce forgetting and show how the robotic system can improve even when switching between different environments.
1 code implementation • CVPR 2021 • Giancarlo Di Biase, Hermann Blum, Roland Siegwart, Cesar Cadena
The inability of state-of-the-art semantic segmentation methods to detect anomaly instances hinders them from being deployed in safety-critical and complex applications, such as autonomous driving.
Ranked #2 on Anomaly Detection on Lost and Found (using extra training data)
no code implementations • 5 Dec 2020 • Janis Postels, Hermann Blum, Yannick Strümpler, Cesar Cadena, Roland Siegwart, Luc van Gool, Federico Tombari
We find that this leads to improved OOD detection of epistemic uncertainty at the cost of ambiguous calibration close to the data distribution.
1 code implementation • 4 Nov 2020 • Le Chen, Yunke Ao, Florian Tschopp, Andrei Cramariuc, Michel Breyer, Jen Jen Chung, Roland Siegwart, Cesar Cadena
Visual-inertial systems rely on precise calibrations of both camera intrinsics and inter-sensor extrinsics, which typically require manually performing complex motions in front of a calibration target.
no code implementations • 3 Nov 2020 • Julia Nitsch, Masha Itkina, Ransalu Senanayake, Juan Nieto, Max Schmidt, Roland Siegwart, Mykel J. Kochenderfer, Cesar Cadena
A mechanism to detect OOD samples is important for safety-critical applications, such as automotive perception, to trigger a safe fallback mode.
no code implementations • 19 Oct 2020 • Alexander Millane, Helen Oleynikova, Christian Lanegger, Jeff Delmerico, Juan Nieto, Roland Siegwart, Marc Pollefeys, Cesar Cadena
Localization of a robotic system within a previously mapped environment is important for reducing estimation drift and for reusing previously built maps.
1 code implementation • 15 Oct 2020 • Berta Bescos, Cesar Cadena, Jose Neira
The first challenge is addressed by the use of a convolutional network that learns a multi-class semantic segmentation of the image.
no code implementations • 10 Aug 2020 • Timo Hinzmann, Tobias Stegemann, Cesar Cadena, Roland Siegwart
In this paper, we present our deep learning-based human detection system that uses optical (RGB) and long-wave infrared (LWIR) cameras to detect, track, localize, and re-identify humans from UAVs flying at high altitude.
1 code implementation • 24 May 2020 • Andrei Cramariuc, Aleksandar Petrov, Rohit Suri, Mayank Mittal, Roland Siegwart, Cesar Cadena
Self-diagnosis and self-repair are some of the key challenges in deploying robotic platforms for long-term real-world applications.
no code implementations • 17 Feb 2020 • Yu Liu, Jie Li, Qingsen Yan, Xia Yuan, Chunxia Zhao, Ian Reid, Cesar Cadena
This paper tackles the problem of data fusion in the semantic scene completion (SSC) task, which can simultaneously deal with semantic labeling and scene completion.
Ranked #14 on 3D Semantic Scene Completion on NYUv2
1 code implementation • 29 Jan 2020 • Yu Liu, Jie Li, Xia Yuan, Chunxia Zhao, Roland Siegwart, Ian Reid, Cesar Cadena
We propose PALNet, a novel hybrid network for SSC based on single depth.
2 code implementations • 5 Dec 2019 • Florian Tschopp, Michael Riner, Marius Fehr, Lukas Bernreiter, Fadri Furrer, Tonci Novkovic, Andreas Pfrunder, Cesar Cadena, Roland Siegwart, Juan Nieto
Robust and accurate pose estimation is crucial for many applications in mobile robotics.
2 code implementations • 27 Sep 2019 • Renaud Dubé, Andrei Cramariuc, Daniel Dugas, Hannes Sommer, Marcin Dymczyk, Juan Nieto, Roland Siegwart, Cesar Cadena
We therefore present SegMap: a map representation solution for localization and mapping based on the extraction of segments in 3D point clouds.
no code implementations • 2 Sep 2019 • David Haldimann, Hermann Blum, Roland Siegwart, Cesar Cadena
There has been a remarkable progress in the accuracy of semantic segmentation due to the capabilities of deep learning.
no code implementations • 1 Aug 2019 • Nicolas Marchal, Charlotte Moraldo, Roland Siegwart, Hermann Blum, Cesar Cadena, Abel Gawel
Foreground objects are therefore detected as areas in an image for which the descriptors are unlikely given the background distribution.
1 code implementation • 23 Jun 2019 • Luigi Freda, Mario Gianni, Fiora Pirri, Abel Gawel, Renaud Dube, Roland Siegwart, Cesar Cadena
This target selection relies on a shared idleness representation and a coordination mechanism preventing topological conflicts.
1 code implementation • 5 Apr 2019 • Hermann Blum, Paul-Edouard Sarlin, Juan Nieto, Roland Siegwart, Cesar Cadena
Deep learning has enabled impressive progress in the accuracy of semantic segmentation.
Ranked #9 on Anomaly Detection on Fishyscapes L&F (using extra training data)
no code implementations • 19 Mar 2019 • Lukas Schaupp, Mathias Bürki, Renaud Dubé, Roland Siegwart, Cesar Cadena
We introduce a novel method for oriented place recognition with 3D LiDAR scans.
1 code implementation • IEEE ROBOTICS AND AUTOMATION LETTERS 2019 • Margarita Grinvald, Fadri Furrer, Tonci Novkovic, Jen Jen Chung, Cesar Cadena, Roland Siegwart, Juan Nieto
To autonomously navigate and plan interactions in real-world environments, robots require the ability to robustly perceive and map complex, unstructured surrounding scenes.
no code implementations • 12 Feb 2019 • Mathias Bürki, Lukas Schaupp, Marcin Dymczyk, Renaud Dubé, Cesar Cadena, Roland Siegwart, Juan Nieto
Changes in appearance is one of the main sources of failure in visual localization systems in outdoor environments.
3 code implementations • CVPR 2019 • Paul-Edouard Sarlin, Cesar Cadena, Roland Siegwart, Marcin Dymczyk
In this paper we propose HF-Net, a hierarchical localization approach based on a monolithic CNN that simultaneously predicts local features and global descriptors for accurate 6-DoF localization.
Ranked #3 on Visual Place Recognition on Berlin Kudamm
1 code implementation • 20 Sep 2018 • Berta Bescos, José Neira, Roland Siegwart, Cesar Cadena
In this paper we present an end-to-end deep learning framework to turn images that show dynamic content, such as vehicles or pedestrians, into realistic static frames.
1 code implementation • 4 Sep 2018 • Paul-Edouard Sarlin, Frédéric Debraine, Marcin Dymczyk, Roland Siegwart, Cesar Cadena
Many robotics applications require precise pose estimates despite operating in large and changing environments.
1 code implementation • 30 Jul 2018 • Hermann Blum, Abel Gawel, Roland Siegwart, Cesar Cadena
Sensor fusion is a fundamental process in robotic systems as it extends the perceptual range and increases robustness in real-world operations.
1 code implementation • 25 Apr 2018 • Renaud Dubé, Andrei Cramariuc, Daniel Dugas, Juan Nieto, Roland Siegwart, Cesar Cadena
While current methods extract descriptors for the single task of localization, SegMap leverages a data-driven descriptor in order to extract meaningful features that can also be used for reconstructing a dense 3D map of the environment and for extracting semantic information.
no code implementations • 28 Sep 2017 • Abel Gawel, Carlo Del Don, Roland Siegwart, Juan Nieto, Cesar Cadena
Our findings show that X-View is able to globally localize aerial-to-ground, and ground-to-ground robot data of drastically different view-points.
no code implementations • 25 Sep 2017 • Mark Pfeiffer, Giuseppe Paolo, Hannes Sommer, Juan Nieto, Roland Siegwart, Cesar Cadena
This paper reports on a data-driven, interaction-aware motion prediction approach for pedestrians in environments cluttered with static obstacles.
2 code implementations • 25 Sep 2016 • Renaud Dubé, Daniel Dugas, Elena Stumm, Juan Nieto, Roland Siegwart, Cesar Cadena
We propose SegMatch, a reliable loop-closure detection algorithm based on the matching of 3D segments.
2 code implementations • 19 Jun 2016 • Cesar Cadena, Luca Carlone, Henry Carrillo, Yasir Latif, Davide Scaramuzza, Jose Neira, Ian Reid, John J. Leonard
Simultaneous Localization and Mapping (SLAM)consists in the concurrent construction of a model of the environment (the map), and the estimation of the state of the robot moving within it.