arXiv subjects

Danica Kragic

Publications and source records attributed to Danica Kragic.

At least 19 recordsLinked to original sources

arXiv2026

Direction-Scale Decomposition in Action Representation: Rethinking What to Tokenize for Vision-Language-Action Models

Action representation plays a central role in discrete-token vision-language-action (VLA) learning but remains underexamined. Under conventional pose-increment representations, action tokens are sensitive to execution speed and dataset-specific normalization, potentially obscuring geometric structure shared across demonstrations and datasets. We introduce Direction-Scale Decomposition (DSD), an action representation that decomposes translation and rotation increments into direction and scale components before tokenization. DSD isolates motion direction while retaining magnitudes in separate scale channels. We evaluate DSD with uniform binning (BIN) and BEAST, a B-spline-based tokenizer, in simulation and real-world manipulation under both single-dataset and mixed-dataset training. On LIBERO, DSD improves average success rates with both tokenizers. On SimplerEnv, DSD-BIN outperforms BIN by 10.3 percentage points in overall success rate under mixed-dataset training. Real-robot experiments further show gains both with and without robotics pretraining. These results support DSD as an effective action representation for discrete-token VLA models and suggest its potential to mitigate performance degradation when training on large and diverse dataset mixtures. Our project page with additional resources is available at https://vla-dsd.github.io/

cs.CV↗

arXiv2026

Hybrid Imitation Learning: Teleoperation Augmentation Primitives that Policies Learn to Trigger

What an operator can demonstrate bounds what imitation learning can learn. Teleoperation interfaces map the human body to the robot, so motions that are hard for a human, such as holding an exact orientation, returning to the same viewpoint, or turning a wrist joint several full revolutions, are difficult to demonstrate on most teleoperation interfaces, even when they are trivial for the robot. We introduce Teleoperation Augmentation Primitives (TAPs): axis locks, perching waypoints, pose anchors, and embodiment-specific routines that the operator triggers during a demonstration by speech, an AR menu, or simply with a controller button. TAPs are themselves recorded in the demonstration and can therefore also be learned and invoked by the policy itself. In simulation, where benchmarks provide human demonstrations, we show that augmenting them with TAPs after the fact yields better policies on a wristcamera-only peg insertion (0.372 to 0.531 success) and on a mug-cleanup task with a learned "remember this pose" anchor (0.203 to 0.323). On a real robot, three tasks show the same pattern: primitives that help the operator do not hurt the policy, and a policy triggering an unscrewing routine succeeds where a plain policy cannot resolve the multi-turn motion from images alone (67% vs. 38% success). We close with an industrial proof-of-concept case leveraging this idea. Project website: https://hybrid-imitation.github.io/

cs.RO↗

arXiv2026

Geometric organization of olfactory descriptor data in the Poincaré disk

Odor quality is commonly represented using high dimensional descriptor profiles, yet their low dimensional organization remains unclear. We investigated whether a two-dimensional hyperbolic embedding can provide an interpretable representation of this structure. We applied hyperbolic metric multidimensional scaling to two complementary datasets: 480 Sagar rating profiles from three participants rating 160 odorants on 15 continuous descriptors, and 4983 GoodScents--Leffingwell molecules annotated with 138 binary descriptors. The embeddings substantially preserved pairwise descriptor distances, supporting subsequent analyses of radial and angular organization. In Sagar, rating profile entropy was strongly and negatively associated with hyperbolic radius, with diffuse profiles closer to the center and concentrated profiles closer to the boundary. This radial organization emerged primarily at the level of the full descriptor profile, rather than any individual descriptor, and remained robust across alternative descriptor representations, participant specific analyses, and averaged ratings. Sweet, musky, fruity, pleasantness showed the strongest directional trends. In GoodScents--Leffingwell, active label entropy, reflecting descriptor multiplicity, increased with radius, whereas orthogonalized descriptor entropy, reflecting spread across orthogonal modes, decreased with radius. Related binary descriptors occupied coherent localized high-density regions. These findings reveal complementary radial and angular organization in the hyperbolic representation of olfactory descriptor data. They support hyperbolic mapping as an interpretable descriptive framework in which radius summarizes global profile properties, while the angular component captures continuous descriptor gradients and categorical organization.

cs.CE↗

arXiv2026

Localized Visual Feature Aggregation via Focus Pooling for Visuomotor Policies

Focusing on spatially localized, control-relevant visual cues has been shown to improve data efficiency in visuomotor policies by reducing the need to model task-irrelevant visual variation. Existing methods often impose this focus through input preprocessing, such as cropping control- or object-centric regions in RGB images or point-clouds. However, it remains underexplored whether such localized features can be exposed directly from commonly used convolutional neural network (CNN) encoded features. In this paper, we show that intermediate CNN features preserve localized visual context for control, but existing pooling methods fail to aggregate it effectively. We introduce FocusPool, an attention pooling module that selectively aggregates intermediate visual features according to their relevance to the robot's current proprioceptive context. The resulting pooled representation captures task-progressive, control-relevant local information and is used directly for policy learning. Across simulation and real-world experiments, FocusPool improves policy success rates over pooling and explicit local focus methods by 36.2% and 41.2%, with training only 5.8% of encoder parameters.

cs.RO↗

arXiv2026

Towards Safe Reinforcement Learning with Reduced Conservativeness: A Case Study on Drone Flight Control

Incorporating formal methods into reinforcement learning (RL) has the potential to result in the best of both worlds, combining the robustness of formal guarantees with the adaptability and learning capabilities of RL, though careful design is needed to balance safety and exploration. In this work, we propose a framework to mitigate this loss of exploration while still allowing for the safety of the system to be ensured. Specifically, we introduce a less restrictive method that can reduce the conservativeness of formal methods by refining a disturbance model using online collected data and it evaluates the safety of a learning-based controller, using computationally efficient zonotopic reachability analysis for the safety analysis to facilitate a real-time implementation. We validate the framework in a real-world drone flight through a canyon, where the drone is subjected to unknown external disturbances and the framework is tasked with learning those disturbances online and adjusting the safety guarantees accordingly. The results show that the framework enables a less restrictive online training of learning-based controllers without compromising the safety of the system.

eess.SY↗

arXiv2026

Attention from Action, for Action: Emergent Visual Bottlenecks for Policy Learning

Visual bottlenecks that focus policy inputs on regions of interest (ROIs) can improve data-efficient visuomotor learning by separating where to look from how to act. Many ROI interfaces rely on external spatial labels, such as gaze, object classes, or affordance annotations. Label-free alternatives often derive crops from trajectories by detecting gripper or motion events and centering a fixed crop at the projected end-effector. Such action-derived crops are useful spatial priors that require no additional labels, but they encode fixed choices about event timing, proxy points, and crop scale. When the visual evidence needed for control lies away from the end-effector or changes continuously with task progress, these crops can become misaligned. We propose Seeker, a task- and state-conditioned readout that learns attention from action. Starting from frozen DINOv3 features, Seeker iteratively updates a query with gathered visual evidence, producing progression-aware ROIs solely from action supervision. The learned ROI serves as a spatial interface for RGB cropping, mask-guided background augmentation, and point-cloud filtering. In simulation and the real world, Seeker improves data efficiency and robustness over no-crop, augmentation, and action-derived crop baselines. On real robots, Seeker raises average in-domain success from the best baseline's 48.3% to 76.7% and success under lighting/background shifts from 20.0% to 60.0%.

cs.RO↗

arXiv2026

Enhancing Visual Domain Robustness in Behaviour Cloning via Saliency-Guided Augmentation

In vision-based behavior cloning (BC), conventional image augmentations such as Random Crop and Color Jitter often fall short under substantial visual domain shifts, including changes in shadows, distractors, and backgrounds. Superimposition-based augmentations, which blend in-domain and out-of-domain images, have shown promise for improving generalization in computer vision, but their suitability for BC remains uncertain because task-critical semantics, spatiotemporal relationships, and agent-target interactions must be preserved. To address this, we introduce RoboSaGA, a Saliency-Guided Augmentation method within the superimposition family tailored for vision-based BC. RoboSaGA dynamically adjusts augmentation intensity at the pixel level using policy-driven saliency, enabling aggressive augmentation in task-irrelevant regions while preserving task-critical information. It integrates seamlessly into existing architectures without requiring structural modifications or additional learning objectives. Experiments in both simulated and real-world settings show that RoboSaGA preserves in-domain performance while substantially improving robustness to visual domain shifts, including distractor and background changes, as well as lighting and shadow variations. Code is available at https://github.com/Zheyu-Zhuang/RoboSaGA.

cs.RO↗

arXiv2026

Deciding When to Rely on Visual Information: Gated Multimodal Fusion in Sequential Recommendation

Multimodal sequential recommender systems commonly fuse visual and collaborative signals uniformly, treating visual features as generically informative regardless of item or user context. We argue that visual utility, defined as the contribution of visual signals to recommendation quality, is a latent contextual variable that depends on both the item and the user's interaction history rather than a fixed item property. To model this variability, we introduce VisGate, a framework that makes adaptive item-level fusion decisions conditioned on item embeddings and the user's current sequence context. Visual representations are learned through a contrastive objective over sequential co-occurrence patterns, preserving complementarity with collaborative embeddings rather than aligning them into a shared space. Beyond achieving competitive recommendation performance, VisGate's learned gate serves as a measurement tool for understanding when and why visual information is beneficial. Our analyses show that visual utility varies across items, increases under interaction sparsity when collaborative signals are weak, and correlates with visual distinctiveness in semantically meaningful ways. Together, these findings highlight the importance of both fine-grained fusion and modality complementarity, while demonstrating that item-level visual utility can be estimated and interpreted through learned gating behaviour.

cs.IR↗

arXiv2026

Learning Physical Interaction: A Survey of Tactile- and Force-aware Robot Learning

Physically grounded robot intelligence requires robots to perceive, reason about, and regulate their interactions with the physical world. This capability is particularly critical in contact-sensitive manipulation, where successful task execution depends not only on visual perception and motion generation, but also on force regulation and adaptive control. In this context, recent robot learning methods have made substantial progress by integrating force, tactile, vision, language, and proprioceptive sensing into learned manipulation policies. In parallel, many systems adopt multi-phase architectures that combine high-level policies, action-refinement modules, and low-level controllers to bridge semantic task understanding with reactive physical execution. Despite these advances, existing surveys have not explicitly reviewed force- and tactile-aware robot learning from a unified perspective that jointly captures multimodal sensing and multi-phase system design. This survey addresses this gap by proposing TF-ART, a Tactile/Force-Aware Robot learning Taxonomy for multimodal and multi-phase frameworks, which maps individual methods into a unified hierarchical structure. The framework characterizes how recent works organize observation modalities, encode and fuse heterogeneous sensory inputs, generate and refine actions across multiple phases, and connect learned policies to reactive robot-end control. Building on this methodological view, we further examine the task settings and infrastructure requirements of physical interaction, thereby integrating both algorithmic and practical perspectives on force- and tactile-aware robot learning.

cs.RO↗

arXiv2026

One Body, Two Minds: Variable Autonomy Approach for a Co-embodied Robotic Hand

Assistive robotic systems face a fundamental trade-off: fully autonomous systems lack user agency, while fully user-controlled systems demand continuous cognitive effort. Existing shared autonomy approaches blend human and robot commands but are mostly deployed in separate physical bodies. We introduce co-embodiment with variable autonomy, where human and robot share a single physical body and operate at different autonomy levels across task phases, from mutual autonomy during object search and grasping to human-dominant control during actuation. We present a co-embodied, wearable robotic hand that has its own ``mind'' and operates with variable autonomy levels. A learning-from-demonstration visuomotor diffusion policy enables autonomous grasping when the user positions the hand near known objects. Once grasped, the system signals completion and the human can actuate the grasped tool (drill, spray bottle, infrared thermometer, lighter, and ice-cream scoop) via hands-free head gestures. The human retains veto authority at all times through a release gesture that returns the system to the initial phase. Unlike blended autonomy, where control is continuously negotiated, our co-embodied approach consists of variable autonomy from full human control to full independent actions while maintaining physical coupling, realizing a one body, two minds paradigm. In a user study with 44 participants performing five bimanual tasks, users rapidly adapted to this ``two minds'' paradigm: completion times improved by 23.3% across trials ($p < 0.001$, Cohen's $d = 0.94$), the best-performing policy variant reached a 93.6% task success rate, and acceptance ratings were high (5.70/7 overall impression, 5.52/7 daily use willingness). This work establishes co-embodiment with variable autonomy as a viable approach for assistive robotics, enabling human-robot collaboration through co-embodiment.

cs.RO↗

arXiv2026

MirrorDuo: Reflection-Consistent Visuomotor Learning from Mirrored Demonstration Pairs

Image-based behaviour cloning leverages demonstrations captured from ubiquitous RGB cameras. However, it remains constrained by the cost of collecting diverse demos, especially for generalizing across workspace variations. We propose MirrorDuo, a reflection-based formulation that operates on image, proprioception, and full 6-DoF end-effector action tuples, generating a mirrored counterpart for each original demonstration, effectively achieving "collect one, get one for free". It can be applied as a data augmentation strategy for existing learning pipelines, such as standard behaviour cloning or diffusion policy, or as a structural prior for reflection-equivariant policy networks. By leveraging the overlap between the original and mirrored domains, MirrorDuo achieves significantly improved performance under the same data budget when demonstrations are evenly distributed across both sides of the workspace. When demonstrations are confined to one side, MirrorDuo enables efficient skill transfer to the mirrored workspace with as few as zero or five demos in the target arrangement.

cs.RO↗

arXiv2026

Equivariant Representation Learning via Class-Pose Decomposition

We introduce a general method for learning representations that are equivariant to symmetries of data. Our central idea is to decompose the latent space into an invariant factor and the symmetry group itself. The components semantically correspond to intrinsic data classes and poses respectively. The learner is trained on a loss encouraging equivariance based on supervision from relative symmetry information. The approach is motivated by theoretical results from group theory and guarantees representations that are lossless, interpretable and disentangled. We provide an empirical investigation via experiments involving datasets with a variety of symmetries. Results show that our representations capture the geometry of data and outperform other equivariant representation learning frameworks.

cs.LG↗

arXiv2026

An Efficient and Continuous Voronoi Density Estimator

We introduce a non-parametric density estimator deemed Radial Voronoi Density Estimator (RVDE). RVDE is grounded in the geometry of Voronoi tessellations and as such benefits from local geometric adaptiveness and broad convergence properties. Due to its radial definition RVDE is continuous and computable in linear time with respect to the dataset size. This amends for the main shortcomings of previously studied VDEs, which are highly discontinuous and computationally expensive. We provide a theoretical study of the modes of RVDE as well as an empirical investigation of its performance on high-dimensional data. Results show that RVDE outperforms other non-parametric density estimators, including recently introduced VDEs.

stat.ME↗

arXiv2026

MoDex: A Diffusion Policy for Sequential Multi-Object Dexterous Grasping

This work addresses sequentially grasping multiple objects with a single dexterous hand without releasing those already held. Most dexterous grasping methods commit all of the hand's degrees of freedom to a single object, underutilizing its dexterity and leaving no redundancy for subsequent grasps. The proposed solution, MoDex, is a diffusion policy that predicts the next gripper pose directly from observations, conditioned on an opposition space and point cloud. The opposition space condition specifies which fingers participate in the current grasp, enabling the gripper to use only a subset of its available degrees of freedom while reserving the remaining degrees of freedom for subsequent grasps. To facilitate sim-to-real transfer, MoDex is trained in two stages: first through imitation learning on expert demonstrations, and subsequently through reinforcement learning fine-tuning, which consistently improves success rates over the pre-trained policy. We evaluate MoDex in simulation on a MuJoCo-based Franka Emika Panda robot equipped with an Allegro Hand and on the corresponding real-world hardware platform. Across both simulation and real-world experiments, MoDex achieves higher success rates than the evaluated learning-based baselines, improving performance by 2.92-17.92% and 6.67-17.78%, respectively. Project page: https://modex2026.github.io/.

cs.RO↗

arXiv2026

Learning Hamiltonian Dynamics at Scale: A Differential-Geometric Approach

Embedding physical intuition into network architectures allows the learning of dynamics that enforce fundamental properties, such as energy conservation laws, thereby leading to physically-plausible predictions. Yet, scaling these models to high-dimensional dynamical systems remains a significant challenge. This paper introduces Reduced-order Hamiltonian Neural Network (RO-HNN), a novel physics-inspired neural network that combines the conservation laws of Hamiltonian mechanics with the scalability of model order reduction. RO-HNN is built on two core components: a novel geometrically-constrained symplectic autoencoder that learns a low-dimensional, structure-preserving symplectic submanifold, and a geometric Hamiltonian neural network that models the dynamics on the submanifold. Our experiments demonstrate that RO-HNN provides physically-consistent, stable, and generalizable predictions of complex high-dimensional dynamics, thereby effectively extending the scope of Hamiltonian neural networks to high-dimensional physical systems.

cs.LG↗

arXiv2026

On the Generalization Capabilities, Design Choices and Limitations of Keypoint Imitation Learning

RGB-based imitation learning requires many demonstrations to generalize to unseen objects or scenes, motivating research into intermediate representations to improve generalization for robotic manipulation. Visual foundation models enable one-shot extraction of keypoints to provide such representation. However, it remains unclear how to integrate them into imitation learning optimally and when they outperform alternative representations. We combine approaches from previous works on keypoint imitation learning (KIL) and investigate several design choices to provide practical guidelines. Using over 2000 real-world rollouts, we also assess the generalization capabilities of KIL to unseen objects and scene variations. KIL achieves a 75% overall success rate across five tasks, significantly outperforming the RGB baseline (47%) and performing on par with S2-diffusion (73%). Finally, we explore the limitations of the foundation models used for keypoint extraction and extend KIL to tasks with multiple object instances. Our results confirm KIL as a data-efficient approach for robot learning, though it does not outperform alternative representations and inherits limitations of the foundation models used for keypoint extraction. All rollout videos, demonstrations, and results are available at https://kil-manipulation.github.io/.

cs.RO↗

arXiv2026

SCENT: Aligning Mass Spectra with Molecular Structure for Olfactory Perception

Predicting human olfactory perception from molecular structure has seen remarkable progress, yet these approaches require explicit chemical structure at inference, which is not available in practical sensing settings. We address this gap by exploring direct electron ionization mass spectrometry (EI-MS), a sensing technique that acquires chemically informative fragmentation fingerprints in seconds, as an alternative input modality for olfactory prediction. We contribute Spectrum-to-Chemical Embedding alignmeNT (SCENT), a multi-modal contrastive learning framework that aligns EI-MS representations with pretrained chemical structure embeddings, while requiring only mass spectra at inference. On the multi-label odor descriptor prediction task, SCENT significantly outperforms MS-only baselines and achieves performance comparable to structure-based models, despite requiring no explicit molecular structure at test time. The learned representations also better approximate continuous human perceptual ratings and generalize to real-world lab-measured spectra, suggesting that cross-modal alignment is an effective strategy for grounding analytical spectra in chemical semantics.

cs.LG↗

arXiv2026

Gesture First, LLM-Assisted Voice Complement: Exploring Multimodal Robot 'Puppeteer' Teleoperation Via Virtual Counterpart in Augmented Reality

Robot teleoperation via augmented reality (AR) offers a promising path toward more intuitive human-robot interaction (HRI). We present a head-mounted AR 'puppeteer' system in which users control a physical robot by interacting with its virtual counterpart robot using large language model (LLM)-assisted voice commands and hand-gesture interaction on the Meta Quest 3. In a within-subject user study with 42 participants performing an AR-based robotic pick-and-place pattern-matching task, we empirically compare two interaction conditions: gesture-only (GO) and combined voice+gesture (VG) on performance and user experience (UX). In VG, voice and gesture operate in a sequential role-allocated manner, with voice handling high-level navigation and gesture handling fine manipulation. Our results show that GO currently provides more reliable and efficient control for this time-critical task, while VG introduces additional flexibility but also latency and recognition issues that can increase workload. We additionally analyze how prior robotics expertise differentiates performance and UX across conditions. Based on these findings, we distill a set of design guidelines for AR 'puppeteer' metaphoric robot teleoperation, framing multimodality as an adaptive strategy that must balance efficiency, robustness, and user expertise rather than assuming that additional modalities are universally beneficial.

cs.HC↗