Papers

Conventional road-situation detection methods achieve strong performance in predefined scenarios but fail in unseen cases and lack semantic interpretation, which is crucial for reliable traffic recommendations. This work introduces a multi-agent AI framework that combines multimodal large language models (MLLMs) with vision-based perception for road-situation monitoring. The framework processes camera feeds and coordinates dedicated agents for situation detection, distance estimation, decision-making, and Cooperative Intelligent Transport System (C-ITS) message generation. Evaluation is conducted on a custom dataset of 103 images extracted from 20 videos of the TAD dataset. Both Gemini-2.0-Flash and Gemini-2.5-Flash were evaluated. The results show 100% recall in situation detection and perfect message schema correctness; however, both models suffer from false-positive detections and have reduced performance in terms of number of lanes, driving lane status and cause code. Surprisingly, Gemini-2.5-Flash, though more capable in general tasks, underperforms Gemini-2.0-Flash in detection accuracy and semantic understanding and incurs higher latency (Table II). These findings motivate further work on fine-tuning specialized LLMs or MLLMs tailored for intelligent transportation applications.

Unmanned Aerial Vehicles (UAVs) are widely used for aerial photography and remote sensing applications. One of the main challenges is to stitch together multiple images into a single high-resolution image that covers a large area. Featurebased image stitching algorithms are commonly used but can suffer from errors and ambiguities in feature detection and matching. To address this, several approaches have been proposed, including using bundle adjustment techniques or direct image alignment. In this paper, we present a novel method that uses a combination of IMU data and computer vision techniques for stitching images captured by a UAV. Our method involves several steps such as estimating the displacement and rotation of the UAV between consecutive images, correcting for perspective distortion, and computing a homography matrix. We then use a standard image stitching algorithm to align and blend the images together. Our proposed method leverages the additional information provided by the IMU data, corrects for various sources of distortion, and can be easily integrated into existing UAV workflows. Our experiments demonstrate the effectiveness and robustness of our method, outperforming some of the existing feature-based image stitching algorithms in terms of accuracy and reliability, particularly in challenging scenarios such as large displacements, rotations, and variations in camera pose.

Zero-shot object navigation (ZSON) in unseen environments remains a challenging problem for household robots, requiring strong perceptual understanding and decision-making capabilities. While recent methods leverage metric maps and Large Language Models (LLMs), they often depend on depth sensors or prebuilt maps, limiting the spatial reasoning ability of Multimodal Large Language Models (MLLMs). Mapless ZSON approaches have emerged to address this, but they typically make short-sighted decisions, leading to local deadlocks due to a lack of historical context. We propose PanoNav, a fully RGB-only, mapless ZSON framework that integrates a Panoramic Scene Parsing module to unlock the spatial parsing potential of MLLMs from panoramic RGB inputs, and a Memory-guided Decision-Making mechanism enhanced by a Dynamic Bounded Memory Queue to incorporate exploration history and avoid local deadlocks. Experiments on the public navigation benchmark show that PanoNav significantly outperforms representative baselines in both SR and SPL metrics.

This paper explores the application of vision-based system identification techniques in quadrotor modeling and control. Through experiments and analysis, we address the complexities and limitations of quadrotor modeling, particularly in relation to thrust and drag coefficients. Grey-box modeling is employed to mitigate uncertainties, and the effectiveness of an onboard vision system is evaluated. An LQR controller is designed based on a system identification model using data from the onboard vision system. The results demonstrate consistent performance between the models, validating the efficacy of vision based system identification. This study highlights the potential of vision-based techniques in enhancing quadrotor modeling and control, contributing to improved performance and operational capabilities. Our findings provide insights into the usability and consistency of these techniques, paving the way for future research in quadrotor performance enhancement, fault detection, and decision-making processes.

The deployment of autonomous service robots in human-centric environments is hindered by a critical gap in perception and planning. Traditional navigation systems rely on expensive LiDARs that, while geometrically precise, are semantically unaware, they cannot distinguish a important document on an office floor from a harmless piece of litter, treating both as physically traversable. While advanced semantic segmentation exists, no prior work has successfully integrated this visual intelligence into a real-time path planner that is efficient enough for low-cost, embedded hardware. This paper presents a framework to bridge this gap, delivering context-aware navigation on an affordable robotic platform. Our approach centers on a novel, tight integration of a lightweight perception module with an online A* planner. The perception system employs a semantic segmentation model to identify user-defined visual constraints, enabling the robot to navigate based on contextual importance rather than physical size alone. This adaptability allows an operator to define what is critical for a given task, be it sensitive papers in an office or safety lines in a factory, thus resolving the ambiguity of what to avoid. This semantic perception is seamlessly fused with geometric data. The identified visual constraints are projected as non-geometric obstacles onto a global map that is continuously updated from sensor data, enabling robust navigation through both partially known and unknown environments. We validate our framework through extensive experiments in high-fidelity simulations and on a real-world robotic platform. The results demonstrate robust, real-time performance, proving that a cost-effective robot can safely navigate complex environments while respecting critical visual cues invisible to traditional planners.

Claims that humanoid robots achieve “human-level” actuation are common but rarely quantified. Peak torque or speed specifications tell us little about whether a joint can deliver the right combination of torque, power, and endurance at task-relevant postures and rates. We introduce a comprehensive framework that makes “human-level” measurable and comparable across systems. Our approach has three components. First, a kinematic DoF atlas standardizes joint coordinate systems and ranges of motion using ISB-based conventions, ensuring that human and robot joints are compared in the same reference frames. Second, Human-Equivalence Envelopes (HEE) define per-joint requirements by measuring whether a robot meets human torque and power simultaneously at the same joint angle and rate $(q,\omega)$, weighted by positive mechanical work in task-specific bands (walking, stairs, lifting, reaching, and hand actions). Third, the Human-Level Actuation Score (HLAS) aggregates six physically grounded factors: workspace coverage (ROM and DoF), HEE coverage, torque-mode bandwidth, efficiency, and thermal sustainability. We provide detailed measurement protocols using dynamometry, electrical power monitoring, and thermal testing that yield every HLAS input from reproducible experiments. A worked example demonstrates HLAS computation for a multi-joint humanoid, showing how the score exposes actuator trade-offs (gearing ratio versus bandwidth and efficiency) that peak-torque specifications obscure. The framework serves as both a design specification for humanoid development and a benchmarking standard for comparing actuation systems, with all components grounded in published human biomechanics data.

Inspired by how humans reason over discrete objects and their relationships, we explore whether compact object-centric and object-relation representations can form a foundation for multitask robotic manipulation. Most existing robotic multitask models rely on dense embeddings that entangle both object and background cues, raising concerns about both efficiency and interpretability. In contrast, we study object-relation-centric representations as a pathway to more structured, efficient, and explainable visuomotor control. Our contributions are two-fold. First, we introduce LIBERO+, a fine-grained benchmark dataset designed to enable and evaluate object-relation reasoning in robotic manipulation. Unlike prior datasets, LIBERO+ provides object-centric annotations that enrich demonstrations with box- and mask-level labels as well as instance-level temporal tracking, supporting compact and interpretable visuomotor representations. Second, we propose SlotVLA, a slot-attention-based framework that captures both objects and their relations for action decoding. It uses a slot-based visual tokenizer to maintain consistent temporal object representations, a relation-centric decoder to produce task-relevant embeddings, and an LLM-driven module that translates these embeddings into executable actions. Experiments on LIBERO+ demonstrate that object-centric slot and object-relation slot representations drastically reduce the number of required visual tokens, while providing competitive generalization. Together, LIBERO+ and SlotVLA provide a compact, interpretable, and effective foundation for advancing object-relation-centric robotic manipulation.

Efficient, accurate, and flexible relative localization is crucial in air-ground collaborative tasks. However, current approaches for robot relative localization are primarily realized in the form of distributed multi-robot SLAM systems with the same sensor configuration, which are tightly coupled with the state estimation of all robots, limiting both flexibility and accuracy. To this end, we fully leverage the high capacity of Unmanned Ground Vehicle (UGV) to integrate multiple sensors, enabling a semi-distributed cross-modal air-ground relative localization framework. In this work, both the UGV and the Unmanned Aerial Vehicle (UAV) independently perform SLAM while extracting deep learning-based keypoints and global descriptors, which decouples the relative localization from the state estimation of all agents. The UGV employs a local Bundle Adjustment (BA) with LiDAR, camera, and an IMU to rapidly obtain accurate relative pose estimates. The BA process adopts sparse keypoint optimization and is divided into two stages: First, optimizing camera poses interpolated from LiDAR-Inertial Odometry (LIO), followed by estimating the relative camera poses between the UGV and UAV. Additionally, we implement an incremental loop closure detection algorithm using deep learning-based descriptors to maintain and retrieve keyframes efficiently. Experimental results demonstrate that our method achieves outstanding performance in both accuracy and efficiency. Unlike traditional multi-robot SLAM approaches that transmit images or point clouds, our method only transmits keypoint pixels and their descriptors, effectively constraining the communication bandwidth under 0.3 Mbps. Codes and data will be publicly available on https://github.com/Ascbpiac/cross-model-relative-localization.git.

Self-supervised goal-conditioned reinforcement learning enables robots to autonomously acquire diverse skills without human supervision. However, a central challenge is the goal setting problem: robots must propose feasible and diverse goals that are achievable in their current environment. Existing methods like RIG (Visual Reinforcement Learning with Imagined Goals) use variational autoencoder (VAE) to generate goals in a learned latent space but have the limitation of producing physically implausible goals that hinder learning efficiency. We propose Physics-Informed RIG (PI-RIG), which integrates physical constraints directly into the VAE training process through a novel Enhanced Physics-Informed Variational Autoencoder (Enhanced p3-VAE), enabling the generation of physically consistent and achievable goals. Our key innovation is the explicit separation of the latent space into physics variables governing object dynamics and environmental factors capturing visual appearance, while enforcing physical consistency through differential equation constraints and conservation laws. This enables the generation of physically consistent and achievable goals that respect fundamental physical principles such as object permanence, collision constraints, and dynamic feasibility. Through extensive experiments, we demonstrate that this physics-informed goal generation significantly improves the quality of proposed goals, leading to more effective exploration and better skill acquisition in visual robotic manipulation tasks including reaching, pushing, and pick-and-place scenarios.

Soft Robotics presents a rich canvas for free-form and continuum devices capable of exerting forces in any direction and transforming between arbitrary configurations. However, there is no current way to tractably and directly exploit the design freedom due to the curse of dimensionality. Parameterisable sets of designs offer a pathway towards tractable, modular soft robotics that appropriately harness the behavioural freeform of soft structures to create rich embodied behaviours. In this work, we present a parametrised class of soft actuators, Programmable Telescopic Soft Pneumatic Actuators (PTSPAs). PTSPAs expand axially on inflation for deployable structures and manipulation in challenging confined spaces. We introduce a parametric geometry generator to customise actuator models from high-level inputs, and explore the new design space through semi-automated experimentation and systematic exploration of key parameters. Using it we characterise the actuators' extension/bending, expansion, and stiffness and reveal clear relationships between key design parameters and performance. Finally we demonstrate the application of the actuators in a deployable soft quadruped whose legs deploy to walk, enabling automatic adaptation to confined spaces. PTSPAs present new design paradigm for deployable and shape morphing structures and wherever large length changes are required.

With the explosive growth of rigid-body simulators, policy learning in simulation has become the de facto standard for most rigid morphologies. In contrast, soft robotic simulation frameworks remain scarce and are seldom adopted by the soft robotics community. This gap stems partly from the lack of easy-to-use, general-purpose frameworks and partly from the high computational cost of accurately simulating continuum mechanics, which often renders policy learning infeasible. In this work, we demonstrate that rapid soft robot policy learning is indeed achievable via implicit time-stepping. Our simulator of choice, DisMech, is a general-purpose, fully implicit soft-body simulator capable of handling both soft dynamics and frictional contact. We further introduce delta natural curvature control, a method analogous to delta joint position control in rigid manipulators, providing an intuitive and effective means of enacting control for soft robot learning. To highlight the benefits of implicit time-stepping and delta curvature control, we conduct extensive comparisons across four diverse soft manipulator tasks against one of the most widely used soft-body frameworks, Elastica. With implicit time-stepping, parallel stepping of 500 environments achieves up to 6x faster speeds for non-contact cases and up to 40x faster for contact-rich scenarios. Finally, a comprehensive sim-to-sim gap evaluation–training policies in one simulator and evaluating them in another–demonstrates that implicit time-stepping provides a rare free lunch: dramatic speedups achieved without sacrificing accuracy.

Vision-language-action (VLA) models hold the promise to attain generalizable embodied control. To achieve this, a pervasive paradigm is to leverage the rich vision-semantic priors of large vision-language models (VLMs). However, the fundamental question persists: How do VLAs effectively inherit the prior knowledge from VLMs? To address this critical question, we introduce a diagnostic benchmark, GrinningFace, an emoji tabletop manipulation task where the robot arm is asked to place objects onto printed emojis corresponding to language instructions. This task design is particularly revealing – knowledge associated with emojis is ubiquitous in Internet-scale datasets used for VLM pre-training, yet emojis themselves are largely absent from standard robotics datasets. Consequently, they provide a clean proxy: successful task completion indicates effective transfer of VLM priors to embodied control. We implement this diagnostic task in both simulated environment and a real robot, and compare various promising techniques for knowledge transfer. Specifically, we investigate the effects of parameter-efficient fine-tuning, VLM freezing, co-training, predicting discretized actions, and predicting latent actions. Through systematic evaluation, our work not only demonstrates the critical importance of preserving VLM priors for the generalization of VLA but also establishes guidelines for future research in developing truly generalizable embodied AI systems.

Accurate measurement of circular centers is a fun-damental geometric sensing problem in instrumentation and measurement tasks involving cameras, LiDARs, and other spa-tial sensors. In circular-target-based LiDAR-camera extrinsic calibration, a 3D circular center measured from LiDAR and its corresponding 2D projected center measured in the image serve as cross-modal geometric observations for estimating the rigid transformation between the two sensor frames. Conven-tional pipelines can bias both measurements: the 3D center is often obtained by sequential plane fitting, point projection, and 2D circle fitting, while the image ellipse center is frequently treated as the projected circle center even though the two generally differ under perspective projection. This paper focuses on accurate 3D and 2D circular-center measurement and uses LiDAR-camera extrinsic calibration as a representative applica-tion. A conformal-geometric-algebra estimator is integrated with RANSAC to jointly recover the 3D center, normal, and radius from noisy or partially observed LiDAR points. A chord-length-variance criterion then estimates the 2D projected center, with its twofold ambiguity resolved by homography validation or a quasi-RANSAC fallback. Synthetic and real-sensor experiments show that the proposed method improves circular-center measurement accuracy and reduces extrinsic calibration error across different target and sensor configurations.

Tendrils coil their shape to anchor the plant to supporting structures, allowing vertical growth toward light. Although climbing plants have been studied for a long time, extracting information regarding the relationship between the temporal shape change, the event that triggers it, and the contact location is still challenging. To help build this relation, we propose an image-based method by which it is possible to analyze shape changes over time in tendrils when mechano-stimulated in different portions of their body. We employ a geometric approach using a 3D Piece-Wise Clothoid-based model to reconstruct the configuration taken by a tendril after mechanical rubbing. The reconstruction shows high robustness and reliability with an accuracy of R2 > 0.99. This method demonstrates distinct advantages over deep learning-based approaches, including reduced data requirements, lower computational costs, and interpretability. Our analysis reveals higher responsiveness in the apical segment of tendrils, which might correspond to higher sensitivity and tissue flexibility in that region of the organs. Our study provides a methodology for gaining new insights into plant biomechanics and offers a foundation for designing and developing novel intelligent robotic systems inspired by climbing plants.

In this paper, we present an underactuated biomimetic underwater robot that is suitable for ecosystem monitoring in both marine and freshwater environments. We present an updated mechanical design for a fish-like robot and propose minimal actuation behaviors learned using reinforcement learning techniques. We present our preliminary mechanical design of the tail oscillation mechanism and illustrate the swimming behaviors on FishGym simulator, where the reinforcement learning techniques will be tested on

Large Language Models (LLMs) have recently emerged as planners for language-instructed agents, generating sequences of actions to accomplish natural language tasks. However, their reliability remains a challenge, especially in long-horizon tasks, since they often produce overconfident yet wrong outputs. Conformal Prediction (CP) has been leveraged to address this issue by wrapping LLM outputs into prediction sets that contain the correct action with a user-defined confidence. When the prediction set is a singleton, the planner executes that action; otherwise, it requests help from a user. This has led to LLM-based planners that can ensure plan correctness with a user-defined probability. However, as LLMs are trained in an uncertainty-agnostic manner, without awareness of prediction sets, they tend to produce unnecessarily large sets, particularly at higher confidence levels, resulting in frequent human interventions limiting autonomous deployment. To address this, we introduce CoFineLLM (Conformal Finetuning for LLMs), the first CP-aware finetuning framework for LLM-based planners that explicitly reduces prediction-set size and, in turn, the need for user interventions. We evaluate our approach on multiple language-instructed robot planning problems and show consistent improvements over uncertainty-aware and uncertainty-agnostic finetuning baselines in terms of prediction-set size, and help rates. Finally, we demonstrate robustness of our method to out-of-distribution scenarios in hardware experiments.

Controlling robots that dynamically engage in contact with their environment is a pressing challenge. Whether a legged robot making-and-breaking contact with a floor, or a manipulator grasping objects, contact is everywhere. Unfortunately, the switching of dynamics at contact boundaries makes control difficult. Predictive controllers face non-convex optimization problems when contact is involved. Here, we overcome this difficulty by applying Koopman operators to subsume the segmented dynamics due to contact changes into a unified, globally-linear model in an embedding space. We show that viscoelastic contact at robot-environment interactions underpins the use of Koopman operators without approximation to control inputs. This methodology enables the convex Model Predictive Control of a legged robot, and the real-time control of a manipulator engaged in dynamic pushing. In this work, we show that our method allows robots to discover elaborate control strategies in real-time over time horizons with multiple contact changes, and the method is applicable to broad fields beyond robotics.

Motivation and gap: PID-family controllers remain a pragmatic choice for many robotic systems due to their simplicity and interpretability, but tuning stable, high-performing gains is time-consuming and typically non-transferable across robot morphologies, payloads, and deployment conditions. Fuzzy gain scheduling can provide interpretable online adjustment, yet its per-joint scaling and consequent parameters are platform-dependent and difficult to tune systematically. Proposed approach: We propose a hierarchical framework for cross-platform tuning of a learnable fuzzy gain-scheduled PID (LF-PID). The controller uses shared fuzzy membership partitions to preserve common error semantics, while learning per-joint scaling and Takagi-Sugeno consequent parameters that schedule PID gains online. Combined with physics-constrained virtual robot synthesis, meta-learning provides cross-platform initialization from robot physical features, and a lightweight reinforcement learning (RL) stage performs deployment-specific refinement under dynamics mismatch. Starting from three base simulated platforms, we generate 232 physically valid training variants via bounded perturbations of mass (+/-10%), inertia (+/-15%), and friction (+/-20%). Results and insight: We evaluate cross-platform generalization on two distinct systems (a 9-DOF serial manipulator and a 12-DOF quadruped) under multiple disturbance scenarios. The RL adaptation stage improves tracking performance on top of the meta-initialized controller, with up to 80.4% error reduction in challenging high-load joints (12.36 degrees to 2.42 degrees) and 19.2% improvement under parameter uncertainty. We further identify an optimization ceiling effect: online refinement yields substantial gains when the meta-initialized baseline exhibits localized deficiencies, but provides limited improvement when baseline quality is already uniformly strong.

Vision Language Models (VLMs) are increasingly used in autonomous driving to help understand traffic scenes, but they sometimes produce hallucinations, which are false details not grounded in the visual input. Detecting and mitigating hallucinations is challenging when ground-truth references are unavailable and model internals are inaccessible. This paper proposes a novel self-contained low-rank approach to automatically rank multiple candidate captions generated by multiple VLMs based on their hallucination levels, using only the captions themselves without requiring external references or model access. By constructing a sentence-embedding matrix and decomposing it into a low-rank consensus component and a sparse residual, we use the residual magnitude to rank captions: selecting the one with the smallest residual as the most hallucination-free. Experiments on the NuScenes dataset demonstrate that our approach achieves 87% selection accuracy in identifying hallucination-free captions, representing a 19% improvement over the unfiltered baseline and a 6-10% improvement over multi-agent debate method. The sorting produced by sparse error magnitudes shows strong correlation with human judgments of hallucinations, validating our scoring mechanism. Additionally, our method, which can be easily parallelized, reduces inference time by 51-67% compared to debate approaches, making it practical for real-time autonomous driving applications.

We study GCS-TSP, a new variant of the Traveling Salesman Problem (TSP) defined over a Graph of Convex Sets (GCS) – a powerful representation for trajectory planning that decomposes the configuration space into convex regions connected by a sparse graph. In this setting, edge costs are not fixed but depend on the specific trajectory selected through each convex region, making classical TSP methods inapplicable. We introduce GHOST, a hierarchical framework that optimally solves the GCS-TSP by combining combinatorial tour search with convex trajectory optimization. GHOST systematically explores tours on a complete graph induced by the GCS, using a novel abstract-path-unfolding algorithm to compute admissible lower bounds that guide best-first search at both the high level (over tours) and the low level (over feasible GCS paths realizing the tour). These bounds provide strong pruning power, enabling efficient search while avoiding unnecessary convex optimization calls. We prove that GHOST guarantees optimality and present a bounded-suboptimal variant for time-critical scenarios. Experiments show that GHOST is orders-of-magnitude faster than unified mixed-integer convex programming baselines for simple cases and uniquely handles complex trajectory planning problems involving high-order continuity constraints and an incomplete GCS.

This chapter addresses the critical challenge of simulation-to-reality (sim-to-real) transfer for deep reinforcement learning (DRL) in bipedal locomotion. After contextualizing the problem within various control architectures, we dissect the “curse of simulation” by analyzing the primary sources of sim-to-real gap: robot dynamics, contact modeling, state estimation, and numerical solvers. Building on this diagnosis, we structure the solutions around two complementary philosophies. The first is to shrink the gap through model-centric strategies that systematically improve the simulator's physical fidelity. The second is to harden the policy, a complementary approach that uses in-simulation robustness training and post-deployment adaptation to make the policy inherently resilient to model inaccuracies. The chapter concludes by synthesizing these philosophies into a strategic framework, providing a clear roadmap for developing and evaluating robust sim-to-real solutions.

While there has been significant progress to use simulated data to learn robotic manipulation of rigid objects, applying its success to deformable objects has been hindered by the lack of both deformable object models and realistic non-rigid body simulators. In this paper, we present Real Garment Benchmark (RGBench), a comprehensive benchmark for robotic manipulation of garments. It features a diverse set of over 6000 garment mesh models, a new high-performance simulator, and a comprehensive protocol to evaluate garment simulation quality with carefully measured real garment dynamics. Our experiments demonstrate that our simulator outperforms currently available cloth simulators by a large margin, reducing simulation error by 20% while maintaining a speed of 3 times faster. We will publicly release RGBench to accelerate future research in robotic garment manipulation. Website: https://rgbench.github.io/

Wheeled bipedal robots have garnered increasing attention in exploration and inspection. However, most research simplifies calculations by ignoring leg dynamics, thereby restricting the robot's full motion potential. Additionally, robots face challenges when traversing uneven terrain. To address the aforementioned issue, we develop a complete dynamics model and design a whole-body control framework with terrain estimation for a novel 6 degrees of freedom wheeled bipedal robot. This model incorporates the closed-loop dynamics of the robot and a ground contact model based on the estimated ground normal vector. We use a LiDAR inertial odometry framework and improved Principal Component Analysis for terrain estimation. Task controllers, including PD control law and LQR, are employed for pose control and centroidal dynamics-based balance control, respectively. Furthermore, a hierarchical optimization approach is used to solve the whole-body control problem. We validate the performance of the terrain estimation algorithm and demonstrate the algorithm's robustness and ability to traverse uneven terrain through both simulation and real-world experiments.

Diffusion policies (DPs) achieve state-of-the-art performance on complex manipulation tasks by learning from large-scale demonstration datasets, often spanning multiple embodiments and environments. However, they cannot guarantee safe behavior, requiring external safety mechanisms. These, however, alter actions in ways unseen during training, causing unpredictable behavior and performance degradation. To address these problems, we propose path-consistent safety filtering (PACS) for DPs. Our approach performs path-consistent braking on a trajectory computed from the sequence of generated actions. In this way, we keep the execution consistent with the training distribution of the policy, maintaining the learned, task-completing behavior. To enable real-time deployment and handle uncertainties, we verify safety using set-based reachability analysis. Our experimental evaluation in simulation and on three challenging real-world human-robot interaction tasks shows that PACS (a) provides formal safety guarantees in dynamic environments, (b) preserves task success rates, and (c) outperforms reactive safety approaches, such as control barrier functions, by up to 68 % in terms of task success. Videos are available at our project website: https://tum-lsy.github.io/pacs.

Robots operating in real-world environments frequently encounter unknown objects with complex structures and articulated components, such as doors, drawers, cabinets, and tools. The ability to perceive, track, and manipulate these objects without prior knowledge of their geometry or kinematic properties remains a fundamental challenge in robotics. In this work, we present a novel method for visuo-tactile-based tracking of unseen objects (single, multiple, or articulated) during robotic interaction without assuming any prior knowledge regarding object shape or dynamics. Our novel pose tracking approach termed ArtReg (stands for Articulated Registration) integrates visuo-tactile point clouds in an unscented Kalman Filter formulation in the SE(3) Lie Group for point cloud registration. ArtReg is used to detect possible articulated joints in objects using purposeful manipulation maneuvers such as pushing or hold-pulling with a two-robot team. Furthermore, we leverage ArtReg to develop a closed-loop controller for goal-driven manipulation of articulated objects to move the object into the desired pose configuration. We have extensively evaluated our approach on various types of unknown objects through real robot experiments. We also demonstrate the robustness of our method by evaluating objects with varying center of mass, low-light conditions, and with challenging visual backgrounds. Furthermore, we benchmarked our approach on a standard dataset of articulated objects and demonstrated improved performance in terms of pose accuracy compared to state-of-the-art methods. Our experiments indicate that robust and accurate pose tracking leveraging visuo-tactile information enables robots to perceive and interact with unseen complex articulated objects (with revolute or prismatic joints).

Humanoid robots are promising to learn a diverse set of human-like locomotion behaviors, including standing up, walking, running, and jumping. However, existing methods predominantly require training independent policies for each skill, yielding behavior-specific controllers that exhibit limited generalization and brittle performance when deployed on irregular terrains and in diverse situations. To address this challenge, we propose Adaptive Humanoid Control (AHC) that adopts a two-stage framework to learn an adaptive humanoid locomotion controller across different skills and terrains. Specifically, we first train several primary locomotion policies and perform a multi-behavior distillation process to obtain a basic multi-behavior controller, facilitating adaptive behavior switching based on the environment. Then, we perform reinforced fine-tuning by collecting online feedback in performing adaptive behaviors on more diverse terrains, enhancing terrain adaptability for the controller. We conduct experiments in both simulation and real-world experiments in Unitree G1 robots. The results show that our method exhibits strong adaptability across various situations and terrains. Project website: https://ahc-humanoid.github.io.

Control barrier functions (CBFs) are a popular tool for safety certification of nonlinear dynamical control systems. Recently, CBFs represented as neural networks have shown great promise due to their expressiveness and applicability to a broad class of dynamics and safety constraints. However, verifying that a trained neural network is indeed a valid CBF is a computational bottleneck that limits the size of the networks that can be used. To overcome this limitation, we present a novel framework for verifying neural CBFs based on piecewise linear upper and lower bounds on the conditions required for a neural network to be a CBF. Our approach is rooted in linear bound propagation (LBP) for neural networks, which we extend to compute bounds on the gradients of the network. Combined with McCormick relaxation, we derive linear upper and lower bounds on the CBF conditions, thereby eliminating the need for computationally expensive verification procedures. Our approach applies to arbitrary control-affine systems and a broad range of nonlinear activation functions. To reduce conservatism, we develop a parallelizable refinement strategy that adaptively refines the regions over which these bounds are computed. Our approach scales to larger neural networks than state-of-the-art verification procedures for CBFs, as demonstrated by our numerical experiments.

We present a tactile sensing method enabled by the mechanical compliance of soft robots; an externally attachable photoreflective module reads surface deformation of silicone skin to estimate contact force without embedding tactile transducers. Locating the sensor off the contact interface reduces damage risk, preserves softness, and simplifies fabrication and maintenance. We first characterize the optical sensing element and the compliant skin, thendetermine the design of a prototype tactile sensor. Compression experiments validate the approach, exhibiting a monotonic force output relationship consistent with theory, low hysteresis, high repeatability over repeated cycles, and small response indentation speeds. We further demonstrate integration on a soft robotic gripper, where the module reliably detects grasp events. Compared with liquid filled or wireembedded tactile skins, the proposed modular add on architecture enhances durability, reduces wiring complexity, and supports straightforward deployment across diverse robot geometries. Because the sensing principle reads skin strain patterns, it also suggests extensions to other somatosensory cues such as joint angle or actuator state estimation from surface deformation. Overall, leveraging surface compliance with an external optical module provides a practical and robust route to equip soft robots with force perception while preserving structural flexibility and manufacturability, paving the way for robotic applications and safe human robot collaboration.

Collision detection is a core component of robotics applications such as simulation, control, and planning. Traditional algorithms like GJK+EPA compute witness points (i.e., the closest or deepest-penetration pairs between two objects) but are inherently non-differentiable, preventing gradient flow and limiting gradient-based optimization in contact-rich tasks such as grasping and manipulation. Recent work introduced efficient first-order randomized smoothing to make witness points differentiable; however, their direction-based formulation is restricted to convex objects and lacks robustness for complex geometries. In this work, we propose a robust and efficient differentiable collision detection framework that supports both convex and concave objects across diverse scales and configurations. Our method introduces distance-based first-order randomized smoothing, adaptive sampling, and equivalent gradient transport for robust and informative gradient computation. Experiments on complex meshes from DexGraspNet and Objaverse show significant improvements over existing baselines. Finally, we demonstrate a direct application of our method for dexterous grasp synthesis to refine the grasp quality. The code is available at https://github.com/JYChen18/DiffCollision.

In open-vocabulary mobile manipulation (OVMM), task success often hinges on the selection of an appropriate base placement for the robot. Existing approaches typically navigate to proximity-based regions without considering affordances, resulting in frequent manipulation failures. We propose Affordance-Guided Coarse-to-Fine Exploration, a zero-shot framework for base placement that integrates semantic understanding from vision-language models (VLMs) with geometric feasibility through an iterative optimization process. Our method constructs cross-modal representations, namely Affordance RGB and Obstacle Map+, to align semantics with spatial context. This enables reasoning that extends beyond the egocentric limitations of RGB perception. To ensure interaction is guided by task-relevant affordances, we leverage coarse semantic priors from VLMs to guide the search toward task-relevant regions and refine placements with geometric constraints, thereby reducing the risk of convergence to local optima. Evaluated on five diverse open-vocabulary mobile manipulation tasks, our system achieves an 85% success rate, significantly outperforming classical geometric planners and VLM-based methods. This demonstrates the promise of affordance-aware and multimodal reasoning for generalizable, instruction-conditioned planning in OVMM.