John Folkesson

Yi Yang, Qingwen Zhang, Ci Li et al.

The evolution of autonomous driving has made remarkable advancements in recent years, evolving into a tangible reality. However, a human-centric large-scale adoption hinges on meeting a variety of multifaceted requirements. To ensure that the autonomous system meets the user's intent, it is essential to accurately discern and interpret user commands, especially in complex or emergency situations. To this end, we propose to leverage the reasoning capabilities of Large Language Models (LLMs) to infer system requirements from in-cabin users' commands. Through a series of experiments that include different LLM models and prompt designs, we explore the few-shot multivariate binary classification accuracy of system requirements from natural language textual commands. We confirm the general ability of LLMs to understand and reason about prompts but underline that their effectiveness is conditioned on the quality of both the LLM model and the design of appropriate sequential prompts. Code and models are public with the link \url{https://github.com/KTH-RPL/DriveCmd_LLM}.

Riccardo Giubilato, Marcus Gerhard Müller, Marco Sewtz et al.

We release the S3LI Vulcano dataset, a multi-modal dataset towards development and benchmarking of Simultaneous Localization and Mapping (SLAM) and place recognition algorithms that rely on visual and LiDAR modalities. Several sequences are recorded on the volcanic island of Vulcano, from the Aeolian Islands in Sicily, Italy. The sequences provide users with data from a variety of environments, textures and terrains, including basaltic or iron-rich rocks, geological formations from old lava channels, as well as dry vegetation and water. The data (rmc.dlr.de/s3li_dataset) is accompanied by an open source toolkit (github.com/DLR-RM/s3li-toolkit) providing tools for generating ground truth poses as well as preparation of labelled samples for place recognition tasks.

Laura Alejandra Encinar Gonzalez, John Folkesson, Rudolph Triebel et al.

Robust loop closure detection is a critical component of Simultaneous Localization and Mapping (SLAM) algorithms in GNSS-denied environments, such as in the context of planetary exploration. In these settings, visual place recognition often fails due to aliasing and weak textures, while LiDAR-based methods suffer from sparsity and ambiguity. This paper presents MPRF, a multimodal pipeline that leverages transformer-based foundation models for both vision and LiDAR modalities to achieve robust loop closure in severely unstructured environments. Unlike prior work limited to retrieval, MPRF integrates a two-stage visual retrieval strategy with explicit 6-DoF pose estimation, combining DINOv2 features with SALAD aggregation for efficient candidate screening and SONATA-based LiDAR descriptors for geometric verification. Experiments on the S3LI dataset and S3LI Vulcano dataset show that MPRF outperforms state-of-the-art retrieval methods in precision while enhancing pose estimation robustness in low-texture regions. By providing interpretable correspondences suitable for SLAM back-ends, MPRF achieves a favorable trade-off between accuracy, efficiency, and reliability, demonstrating the potential of foundation models to unify place recognition and pose estimation. Code and models will be released at github.com/DLR-RM/MPRF.

Yiping Xie, Nils Bore, John Folkesson

Sidescan sonar intensity encodes information about the changes of surface normal of the seabed. However, other factors such as seabed geometry as well as its material composition also affect the return intensity. One can model these intensity changes in a forward direction from the surface normals from bathymetric map and physical properties to the measured intensity or alternatively one can use an inverse model which starts from the intensities and models the surface normals. Here we use an inverse model which leverages deep learning's ability to learn from data; a convolutional neural network is used to estimate the surface normal from the sidescan. Thus the internal properties of the seabed are only implicitly learned. Once this information is estimated, a bathymetric map can be reconstructed through an optimization framework that also includes altimeter readings to provide a sparse depth profile as a constraint. Implicit neural representation learning was recently proposed to represent the bathymetric map in such an optimization framework. In this article, we use a neural network to represent the map and optimize it under constraints of altimeter points and estimated surface normal from sidescan. By fusing multiple observations from different angles from several sidescan lines, the estimated results are improved through optimization. We demonstrate the efficiency and scalability of the approach by reconstructing a high-quality bathymetry using sidescan data from a large sidescan survey. We compare the proposed data-driven inverse model approach of modeling a sidescan with a forward Lambertian model. We assess the quality of each reconstruction by comparing it with data constructed from a multibeam sensor.

Yiping Xie, Nils Bore, John Folkesson

We propose a novel data-driven approach for high-resolution bathymetric reconstruction from sidescan. Sidescan sonar (SSS) intensities as a function of range do contain some information about the slope of the seabed. However, that information must be inferred. Additionally, the navigation system provides the estimated trajectory, and normally the altitude along this trajectory is also available. From these we obtain a very coarse seabed bathymetry as an input. This is then combined with the indirect but high-resolution seabed slope information from the sidescan to estimate the full bathymetry. This sparse depth could be acquired by single-beam echo sounder, Doppler Velocity Log (DVL), other bottom tracking sensors or bottom tracking algorithm from sidescan itself. In our work, a fully convolutional network is used to estimate the depth contour and its aleatoric uncertainty from the sidescan images and sparse depth in an end-to-end fashion. The estimated depth is then used together with the range to calculate the point's 3D location on the seafloor. A high-quality bathymetric map can be reconstructed after fusing the depth predictions and the corresponding confidence measures from the neural networks. We show the improvement of the bathymetric map gained by using sparse depths with sidescan over estimates with sidescan alone. We also show the benefit of confidence weighting when fusing multiple bathymetric estimates into a single map.

Weiqi Xu, Li Ling, Yiping Xie et al.

Acoustic sensors play an important role in autonomous underwater vehicles (AUVs). Sidescan sonar (SSS) detects a wide range and provides photo-realistic images in high resolution. However, SSS projects the 3D seafloor to 2D images, which are distorted by the AUV's altitude, target's range and sensor's resolution. As a result, the same physical area can show significant visual differences in SSS images from different survey lines, causing difficulties in tasks such as pixel correspondence and template matching. In this paper, a canonical transformation method consisting of intensity correction and slant range correction is proposed to decrease the above distortion. The intensity correction includes beam pattern correction and incident angle correction using three different Lambertian laws (cos, cos2, cot), whereas the slant range correction removes the nadir zone and projects the position of SSS elements into equally horizontally spaced, view-point independent bins. The proposed method is evaluated on real data collected by a HUGIN AUV, with manually-annotated pixel correspondence as ground truth reference. Experimental results on patch pairs compare similarity measures and keypoint descriptor matching. The results show that the canonical transformation can improve the patch similarity, as well as SIFT descriptor matching accuracy in different images where the same physical area was ensonified.

Li Ling, Yiping Xie, Nils Bore et al.

Multibeam echo-sounder (MBES) is the de-facto sensor for bathymetry mapping. In recent years, cheaper MBES sensors and global mapping initiatives have led to exponential growth of available data. However, raw MBES data contains 1-25% of noise that requires semi-automatic filtering using tools such as Combined Uncertainty and Bathymetric Estimator (CUBE). In this work, we draw inspirations from the 3D point cloud community and adapted a score-based point cloud denoising network for MBES outlier detection and denoising. We trained and evaluated this network on real MBES survey data. The proposed method was found to outperform classical methods, and can be readily integrated into existing MBES standard workflow. To facilitate future research, the code and pretrained model are available online.

Ignacio Torroba, Marco Chella, Aldo Teran et al.

Autonomous underwater vehicles (AUVs) are becoming standard tools for underwater exploration and seabed mapping in both scientific and industrial applications \cite{graham2022rapid, stenius2022system}. Their capacity to dive untethered allows them to reach areas inaccessible to surface vessels and to collect data more closely to the seafloor, regardless of the water depth. However, their navigation autonomy remains bounded by the accuracy of their dead reckoning (DR) estimate of their global position, severely limited in the absence of a priori maps of the area and GPS signal. Global localization systems equivalent to the later exists for the underwater domain, such as LBL or USBL. However they involve expensive external infrastructure and their reliability decreases with the distance to the AUV, making them unsuitable for deep sea surveys.

Yi Yang, Qingwen Zhang, Thomas Gilles et al.

As the pretraining technique is growing in popularity, little work has been done on pretrained learning-based motion prediction methods in autonomous driving. In this paper, we propose a framework to formalize the pretraining task for trajectory prediction of traffic participants. Within our framework, inspired by the random masked model in natural language processing (NLP) and computer vision (CV), objects' positions at random timesteps are masked and then filled in by the learned neural network (NN). By changing the mask profile, our framework can easily switch among a range of motion-related tasks. We show that our proposed pretraining framework is able to deal with noisy inputs and improves the motion prediction accuracy and miss rate, especially for objects occluded over time by evaluating it on Argoverse and NuScenes datasets.

Aldo Terán Espinoza, Antonio Terán Espinoza, John Folkesson et al.

Due to the challenges regarding the limits of their endurance and autonomous capabilities, underwater docking for autonomous underwater vehicles (AUVs) has become a topic of interest for many academic and commercial applications. Herein, we take on the problem of relative navigation for the generalized version of the docking operation, which we address as proximity operations. Proximity operations typically involve only two actors, a chaser and a target. We leverage the similarities to proximity operations (prox-ops) from spacecraft robotic missions to frame the diverse docking scenarios with a set of phases the chaser undergoes on the way to its target. We emphasize the versatility on the use of factor graphs as a generalized representation to model the underlying simultaneous trajectory estimation and relative navigation (STERN) problem that arises with any prox-ops scenario, regardless of the sensor suite or the agents' dynamic constraints. To emphasize the flexibility of factor graphs as the modeling foundation for arbitrary underwater prox-ops, we compile a list of state-of-the-art research in the field and represent the different scenario using the same factor graph representation. We detail the procedure required to model, design, and implement factor graph-based estimators by addressing a long-distance acoustic homing scenario of an AUV to a moving mothership using datasets from simulated and real-world deployments; an analysis of these results is provided to shed light on the flexibility and limitations of the dynamic assumptions of the moving target. A description of our front- and back-end is also presented together with a timing breakdown of all processes to show its potential deployment on a real-time system.

Ioanna Mitsioni, Joonatan Mänttäri, Yiannis Karayiannidis et al.

Deep Neural Networks (NNs) have been widely utilized in contact-rich manipulation tasks to model the complicated contact dynamics. However, NN-based models are often difficult to decipher which can lead to seemingly inexplicable behaviors and unidentifiable failure cases. In this work, we address the interpretability of NN-based models by introducing the kinodynamic images. We propose a methodology that creates images from the kinematic and dynamic data of a contact-rich manipulation task. Our formulation visually reflects the task's state by encoding its kinodynamic variations and temporal evolution. By using images as the state representation, we enable the application of interpretability modules that were previously limited to vision-based tasks. We use this representation to train Convolution-based Networks and we extract interpretations of the model's decisions with Grad-CAM, a technique that produces visual explanations. Our method is versatile and can be applied to any classification problem using synchronous features in manipulation to visually interpret which parts of the input drive the model's decisions and distinguish its failure modes. We evaluate this approach on two examples of real-world contact-rich manipulation: pushing and cutting, with known and unknown objects. Finally, we demonstrate that our method enables both detailed visual inspections of sequences in a task, as well as high-level evaluations of a model's behavior and tendencies. Data and code for this work are available at https://github.com/imitsioni/interpretable_manipulation.

Yi Yang, Kei Ikemura, Qingwen Zhang et al.

Recent multi-task learning studies suggest that linear scalarization, when using well-chosen fixed task weights, can achieve comparable to or even better performance than complex multi-task optimization (MTO) methods. It remains unclear why certain weights yield optimal performance and how to determine these weights without relying on exhaustive hyperparameter search. This paper establishes a direct connection between linear scalarization and MTO methods, revealing through extensive experiments that well-performing scalarization weights exhibit specific trends in key MTO metrics, such as high gradient magnitude similarity. Building on this insight, we introduce AutoScale, a simple yet effective two-phase framework that uses these MTO metrics to guide weight selection for linear scalarization, without expensive weight search. AutoScale consistently shows superior performance with high efficiency across diverse datasets including a new large-scale benchmark.

Li Ling, Jun Zhang, Nils Bore et al.

Deep learning has shown promising results for multiple 3D point cloud registration datasets. However, in the underwater domain, most registration of multibeam echo-sounder (MBES) point cloud data are still performed using classical methods in the iterative closest point (ICP) family. In this work, we curate and release DotsonEast Dataset, a semi-synthetic MBES registration dataset constructed from an autonomous underwater vehicle in West Antarctica. Using this dataset, we systematically benchmark the performance of 2 classical and 4 learning-based methods. The experimental results show that the learning-based methods work well for coarse alignment, and are better at recovering rough transforms consistently at high overlap (20-50%). In comparison, GICP (a variant of ICP) performs well for fine alignment and is better across all metrics at extremely low overlap (10%). To the best of our knowledge, this is the first work to benchmark both learning-based and classical registration methods on an AUV-based MBES dataset. To facilitate future research, both the code and data are made available online.

Ignacio Torrobam Nils Bore, John Folkesson

Both higher efficiency and cost reduction can be gained from automating bathymetric surveying for offshore applications such as pipeline, telecommunication or power cables installation and inspection on the seabed. We present a SLAM system that optimizes the geo-referencing of bathymetry surveys by fusing the dead-reckoning sensor data from the surveying vehicle with constraints from the maximization of the geometric consistency of overlapping regions of the survey. The framework has been extensively tested on bathymetric maps from both simulation and several actual industrial surveys and has proved robustness over different types of terrain. We demonstrate that our system is able to maximize the consistency of the final map even when there are large sections of the survey with reduced topographic variation. The framework has been made publicly available together with the simulation environment used to test it and some of the datasets.

Ignacio Torroba, Christopher Iliffe Sprague, Nils Bore et al.

Registration methods for point clouds have become a key component of many SLAM systems on autonomous vehicles. However, an accurate estimate of the uncertainty of such registration is a key requirement to a consistent fusion of this kind of measurements in a SLAM filter. This estimate, which is normally given as a covariance in the transformation computed between point cloud reference frames, has been modelled following different approaches, among which the most accurate is considered to be the Monte Carlo method. However, a Monte Carlo approximation is cumbersome to use inside a time-critical application such as online SLAM. Efforts have been made to estimate this covariance via machine learning using carefully designed features to abstract the raw point clouds. However, the performance of this approach is sensitive to the features chosen. We argue that it is possible to learn the features along with the covariance by working with the raw data and thus we propose a new approach based on PointNet. In this work, we train this network using the KL divergence between the learned uncertainty distribution and one computed by the Monte Carlo method as the loss. We test the performance of the general model presented applying it to our target use-case of SLAM with an autonomous underwater vehicle (AUV) restricted to the 2-dimensional registration of 3D bathymetric point clouds.

Joonatan Mänttäri, Sofia Broomé, John Folkesson et al.

A number of techniques for interpretability have been presented for deep learning in computer vision, typically with the goal of understanding what the networks have based their classification on. However, interpretability for deep video architectures is still in its infancy and we do not yet have a clear concept of how to decode spatiotemporal features. In this paper, we present a study comparing how 3D convolutional networks and convolutional LSTM networks learn features across temporally dependent frames. This is the first comparison of two video models that both convolve to learn spatial features but have principally different methods of modeling time. Additionally, we extend the concept of meaningful perturbation introduced by \cite{MeaningFulPert} to the temporal dimension, to identify the temporal part of a sequence most meaningful to the network for a classification decision. Our findings indicate that the 3D convolutional model concentrates on shorter events in the input sequence, and places its spatial focus on fewer, contiguous areas.

Jiexiong Tang, John Folkesson, Patric Jensfelt

In this paper, we proposed a new deep learning based dense monocular SLAM method. Compared to existing methods, the proposed framework constructs a dense 3D model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner.Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.

Jiexiong Tang, Ludvig Ericson, John Folkesson et al.

In this paper, we present a deep learning-based network, GCNv2, for generation of keypoints and descriptors. GCNv2 is built on our previous method, GCN, a network trained for 3D projective geometry. GCNv2 is designed with a binary descriptor vector as the ORB feature so that it can easily replace ORB in systems such as ORB-SLAM2. GCNv2 significantly improves the computational efficiency over GCN that was only able to run on desktop hardware. We show how a modified version of ORB-SLAM2 using GCNv2 features runs on a Jetson TX2, an embedded low-power platform. Experimental results show that GCNv2 retains comparable accuracy as GCN and that it is robust enough to use for control of a flying drone.

Xi Chen, Ali Ghadirzadeh, John Folkesson et al.

Mobile robot navigation in complex and dynamic environments is a challenging but important problem. Reinforcement learning approaches fail to solve these tasks efficiently due to reward sparsities, temporal complexities and high-dimensionality of sensorimotor spaces which are inherent in such problems. We present a novel approach to train action policies to acquire navigation skills for wheel-legged robots using deep reinforcement learning. The policy maps height-map image observations to motor commands to navigate to a target position while avoiding obstacles. We propose to acquire the multifaceted navigation skill by learning and exploiting a number of manageable navigation behaviors. We also introduce a domain randomization technique to improve the versatility of the training samples. We demonstrate experimentally a significant improvement in terms of data-efficiency, success rate, robustness against irrelevant sensory data, and also the quality of the maneuver skills.

Nils Bore, Patric Jensfelt, John Folkesson

In this work, we present a method for tracking and learning the dynamics of all objects in a large scale robot environment. A mobile robot patrols the environment and visits the different locations one by one. Movable objects are discovered by change detection, and tracked throughout the robot deployment. For tracking, we extend the Rao-Blackwellized particle filter of previous work with birth and death processes, enabling the method to handle an arbitrary number of objects. Target births and associations are sampled using Gibbs sampling. The parameters of the system are then learnt using the Expectation Maximization algorithm in an unsupervised fashion. The system therefore enables learning of the dynamics of one particular environment, and of its objects. The algorithm is evaluated on data collected autonomously by a mobile robot in an office environment during a real-world deployment. We show that the algorithm automatically identifies and tracks the moving objects within 3D maps and infers plausible dynamics models, significantly decreasing the modeling bias of our previous work. The proposed method represents an improvement over previous methods for environment dynamics learning as it allows for learning of fine grained processes.

Nils Bore, Johan Ekekrantz, Patric Jensfelt et al.

This paper studies the problem of detection and tracking of general objects with long-term dynamics, observed by a mobile robot moving in a large environment. A key problem is that due to the environment scale, it can only observe a subset of the objects at any given time. Since some time passes between observations of objects in different places, the objects might be moved when the robot is not there. We propose a model for this movement in which the objects typically only move locally, but with some small probability they jump longer distances, through what we call global motion. For filtering, we decompose the posterior over local and global movements into two linked processes. The posterior over the global movements and measurement associations is sampled, while we track the local movement analytically using Kalman filters. This novel filter is evaluated on point cloud data gathered autonomously by a mobile robot over an extended period of time. We show that tracking jumping objects is feasible, and that the proposed probabilistic treatment outperforms previous methods when applied to real world data. The key to efficient probabilistic tracking in this scenario is focused sampling of the object posteriors.

Johan Ekekrantz, Nils Bore, Rares Ambrus et al.

In this paper we introduce a system for unsupervised object discovery and segmentation of RGBD-images. The system models the sensor noise directly from data, allowing accurate segmentation without sensor specific hand tuning of measurement noise models making use of the recently introduced Statistical Inlier Estimation (SIE) method. Through a fully probabilistic formulation, the system is able to apply probabilistic inference, enabling reliable segmentation in previously challenging scenarios. In addition, we introduce new methods for filtering out false positives, significantly improving the signal to noise ratio. We show that the system significantly outperform state-of-the-art in on a challenging real-world dataset.

Johan Ekekrantz, John Folkesson, Patric Jensfelt

In this paper we introduce an adaptive cost function for pointcloud registration. The algorithm automatically estimates the sensor noise, which is important for generalization across different sensors and environments. Through experiments on real and synthetic data, we show significant improvements in accuracy and robustness over state-of-the-art solutions.

Nick Hawes, Chris Burbridge, Ferdian Jovan et al.

Thanks to the efforts of the robotics and autonomous systems community, robots are becoming ever more capable. There is also an increasing demand from end-users for autonomous service robots that can operate in real environments for extended periods. In the STRANDS project we are tackling this demand head-on by integrating state-of-the-art artificial intelligence and robotics research into mobile service robots, and deploying these systems for long-term installations in security and care environments. Over four deployments, our robots have been operational for a combined duration of 104 days autonomously performing end-user defined tasks, covering 116km in the process. In this article we describe the approach we have used to enable long-term autonomous operation in everyday environments, and how our robots are able to use their long run times to improve their own performance.