Robotics 45
☆ Drone Swarm Energy Management
This note presents an analytical framework for decision-making in drone swarm systems operating under uncertainty, based on the integration of Partially Observable Markov Decision Processes (POMDP) with Deep Deterministic Policy Gradient (DDPG) reinforcement learning. The proposed approach enables adaptive control and cooperative behavior of unmanned aerial vehicles (UAVs) within a cognitive AI platform, where each agent learns optimal energy management and navigation policies from dynamic environmental states. We extend the standard DDPG architecture with a belief-state representation derived from Bayesian filtering, allowing for robust decision-making in partially observable environments. In this paper, for the Gaussian case, we numerically compare the performance of policies derived from DDPG to optimal policies for discretized versions of the original continuous problem. Simulation results demonstrate that the POMDP-DDPG-based swarm control model significantly improves mission success rates and energy efficiency compared to baseline methods. The developed framework supports distributed learning and decision coordination across multiple agents, providing a foundation for scalable cognitive swarm autonomy. The outcomes of this research contribute to the advancement of energy-aware control algorithms for intelligent multi-agent systems and can be applied in security, environmental monitoring, and infrastructure inspection scenarios.
comment: 14 pages, 4 Tables, 2 Figures
☆ Volumetric Ergodic Control
Ergodic control synthesizes optimal coverage behaviors over spatial distributions for nonlinear systems. However, existing formulations model the robot as a non-volumetric point, but in practice a robot interacts with the environment through its body and sensors with physical volume. In this work, we introduce a new ergodic control formulation that optimizes spatial coverage using a volumetric state representation. Our method preserves the asymptotic coverage guarantees of ergodic control, adds minimal computational overhead for real-time control, and supports arbitrary sample-based volumetric models. We evaluate our method across search and manipulation tasks -- with multiple robot dynamics and end-effector geometries or sensor models -- and show that it improves coverage efficiency by more than a factor of two while maintaining a 100% task completion rate across all experiments, outperforming the standard ergodic control method. Finally, we demonstrate the effectiveness of our method on a robot arm performing mechanical erasing tasks.
comment: 8 pages, 8 figures
☆ Terrain Costmap Generation via Scaled Preference Conditioning
Successful autonomous robot navigation in off-road domains requires the ability to generate high-quality terrain costmaps that are able to both generalize well over a wide variety of terrains and rapidly adapt relative costs at test time to meet mission-specific needs. Existing approaches for costmap generation allow for either rapid test-time adaptation of relative costs (e.g., semantic segmentation methods) or generalization to new terrain types (e.g., representation learning methods), but not both. In this work, we present scaled preference conditioned all-terrain costmap generation (SPACER), a novel approach for generating terrain costmaps that leverages synthetic data during training in order to generalize well to new terrains, and allows for rapid test-time adaptation of relative costs by conditioning on a user-specified scaled preference context. Using large-scale aerial maps, we provide empirical evidence that SPACER outperforms other approaches at generating costmaps for terrain navigation, with the lowest measured regret across varied preferences in five of seven environments for global path planning.
☆ Scalable Policy Evaluation with Video World Models
Training generalist policies for robotic manipulation has shown great promise, as they enable language-conditioned, multi-task behaviors across diverse scenarios. However, evaluating these policies remains difficult because real-world testing is expensive, time-consuming, and labor-intensive. It also requires frequent environment resets and carries safety risks when deploying unproven policies on physical robots. Manually creating and populating simulation environments with assets for robotic manipulation has not addressed these issues, primarily due to the significant engineering effort required and the often substantial sim-to-real gap, both in terms of physics and rendering. In this paper, we explore the use of action-conditional video generation models as a scalable way to learn world models for policy evaluation. We demonstrate how to incorporate action conditioning into existing pre-trained video generation models. This allows leveraging internet-scale in-the-wild online videos during the pre-training stage, and alleviates the need for a large dataset of paired video-action data, which is expensive to collect for robotic manipulation. Our paper examines the effect of dataset diversity, pre-trained weight and common failure cases for the proposed evaluation pipeline.Our experiments demonstrate that, across various metrics, including policy ranking and the correlation between actual policy values and predicted policy values, these models offer a promising approach for evaluating policies without requiring real-world interactions.
☆ Scalable Coverage Trajectory Synthesis on GPUs as Statistical Inference
Coverage motion planning is essential to a wide range of robotic tasks. Unlike conventional motion planning problems, which reason over temporal sequences of states, coverage motion planning requires reasoning over the spatial distribution of entire trajectories, making standard motion planning methods limited in computational efficiency and less amenable to modern parallelization frameworks. In this work, we formulate the coverage motion planning problem as a statistical inference problem from the perspective of flow matching, a generative modeling technique that has gained significant attention in recent years. The proposed formulation unifies commonly used statistical discrepancy measures, such as Kullback-Leibler divergence and Sinkhorn divergence, with a standard linear quadratic regulator problem. More importantly, it decouples the generation of trajectory gradients for coverage from the synthesis of control under nonlinear system dynamics, enabling significant acceleration through parallelization on modern computational architectures, particularly Graphics Processing Units (GPUs). This paper focuses on the advantages of this formulation in terms of scalability through parallelization, highlighting its computational benefits compared to conventional methods based on waypoint tracking.
comment: Presented at the "Workshop on Fast Motion Planning and Control in the Era of Parallelism" at Robotics: Science and Systems 2025. Workshop website: https://sites.google.com/rice.edu/parallelized-planning-control/
☆ Collaborative Representation Learning for Alignment of Tactile, Language, and Vision Modalities
Tactile sensing offers rich and complementary information to vision and language, enabling robots to perceive fine-grained object properties. However, existing tactile sensors lack standardization, leading to redundant features that hinder cross-sensor generalization. Moreover, existing methods fail to fully integrate the intermediate communication among tactile, language, and vision modalities. To address this, we propose TLV-CoRe, a CLIP-based Tactile-Language-Vision Collaborative Representation learning method. TLV-CoRe introduces a Sensor-Aware Modulator to unify tactile features across different sensors and employs tactile-irrelevant decoupled learning to disentangle irrelevant tactile features. Additionally, a Unified Bridging Adapter is introduced to enhance tri-modal interaction within the shared representation space. To fairly evaluate the effectiveness of tactile models, we further propose the RSS evaluation framework, focusing on Robustness, Synergy, and Stability across different methods. Experimental results demonstrate that TLV-CoRe significantly improves sensor-agnostic representation learning and cross-modal alignment, offering a new direction for multimodal tactile representation.
☆ A Comparative Evaluation of Prominent Methods in Autonomous Vehicle Certification
The "Vision Zero" policy, introduced by the Swedish Parliament in 1997, aims to eliminate fatalities and serious injuries resulting from traffic accidents. To achieve this goal, the use of self-driving vehicles in traffic is envisioned and a roadmap for the certification of self-driving vehicles is aimed to be determined. However, it is still unclear how the basic safety requirements that autonomous vehicles must meet will be verified and certified, and which methods will be used. This paper focuses on the comparative evaluation of the prominent methods planned to be used in the certification process of autonomous vehicles. It examines the prominent methods used in the certification process, develops a pipeline for the certification process of autonomous vehicles, and determines the stages, actors, and areas where the addressed methods can be applied.
☆ Rethinking Progression of Memory State in Robotic Manipulation: An Object-Centric Perspective AAAI 2026
Nhat Chung, Taisei Hanyu, Toan Nguyen, Huy Le, Frederick Bumgarner, Duy Minh Ho Nguyen, Khoa Vo, Kashu Yamazaki, Chase Rainwater, Tung Kieu, Anh Nguyen, Ngan Le
As embodied agents operate in increasingly complex environments, the ability to perceive, track, and reason about individual object instances over time becomes essential, especially in tasks requiring sequenced interactions with visually similar objects. In these non-Markovian settings, key decision cues are often hidden in object-specific histories rather than the current scene. Without persistent memory of prior interactions (what has been interacted with, where it has been, or how it has changed) visuomotor policies may fail, repeat past actions, or overlook completed ones. To surface this challenge, we introduce LIBERO-Mem, a non-Markovian task suite for stress-testing robotic manipulation under object-level partial observability. It combines short- and long-horizon object tracking with temporally sequenced subgoals, requiring reasoning beyond the current frame. However, vision-language-action (VLA) models often struggle in such settings, with token scaling quickly becoming intractable even for tasks spanning just a few hundred frames. We propose Embodied-SlotSSM, a slot-centric VLA framework built for temporal scalability. It maintains spatio-temporally consistent slot identities and leverages them through two mechanisms: (1) slot-state-space modeling for reconstructing short-term history, and (2) a relational encoder to align the input tokens with action decoding. Together, these components enable temporally grounded, context-aware action prediction. Experiments show Embodied-SlotSSM's baseline performance on LIBERO-Mem and general tasks, offering a scalable solution for non-Markovian reasoning in object-centric robotic policies.
comment: Accepted at AAAI 2026
☆ SimTac: A Physics-Based Simulator for Vision-Based Tactile Sensing with Biomorphic Structures
Xuyang Zhang, Jiaqi Jiang, Zhuo Chen, Yongqiang Zhao, Tianqi Yang, Daniel Fernandes Gomes, Jianan Wang, Shan Luo
Tactile sensing in biological organisms is deeply intertwined with morphological form, such as human fingers, cat paws, and elephant trunks, which enables rich and adaptive interactions through a variety of geometrically complex structures. In contrast, vision-based tactile sensors in robotics have been limited to simple planar geometries, with biomorphic designs remaining underexplored. To address this gap, we present SimTac, a physics-based simulation framework for the design and validation of biomorphic tactile sensors. SimTac consists of particle-based deformation modeling, light-field rendering for photorealistic tactile image generation, and a neural network for predicting mechanical responses, enabling accurate and efficient simulation across a wide range of geometries and materials. We demonstrate the versatility of SimTac by designing and validating physical sensor prototypes inspired by biological tactile structures and further demonstrate its effectiveness across multiple Sim2Real tactile tasks, including object classification, slip detection, and contact safety assessment. Our framework bridges the gap between bio-inspired design and practical realisation, expanding the design space of tactile sensors and paving the way for tactile sensing systems that integrate morphology and sensing to enable robust interaction in unstructured environments.
☆ RadAround: A Field-Expedient Direction Finder for Contested IoT Sensing & EM Situational Awareness
This paper presents RadAround, a passive 2-D direction-finding system designed for adversarial IoT sensing in contested environments. Using mechanically steered narrow-beam antennas and field-deployable SCADA software, it generates high-resolution electromagnetic (EM) heatmaps using low-cost COTS or 3D-printed components. The microcontroller-deployable SCADA coordinates antenna positioning and SDR sampling in real time for resilient, on-site operation. Its modular design enables rapid adaptation for applications such as EMC testing in disaster-response deployments, battlefield spectrum monitoring, electronic intrusion detection, and tactical EM situational awareness (EMSA). Experiments show RadAround detecting computing machinery through walls, assessing utilization, and pinpointing EM interference (EMI) leakage sources from Faraday enclosures.
comment: 6 pages. Cite as O. Maute, B. A. Roberts, and B. Peköz, "RadAround: A field-expedient direction finder for contested IoT sensing & EM situational awareness," in Proc. 2025 IEEE Military Commun. Conf. (MILCOM), Los Angeles, USA, Oct. 2025, pp. 1-6
☆ Simulating an Autonomous System in CARLA using ROS 2
Autonomous racing offers a rigorous setting to stress test perception, planning, and control under high speed and uncertainty. This paper proposes an approach to design and evaluate a software stack for an autonomous race car in CARLA: Car Learning to Act simulator, targeting competitive driving performance in the Formula Student UK Driverless (FS-AI) 2025 competition. By utilizing a 360° light detection and ranging (LiDAR), stereo camera, global navigation satellite system (GNSS), and inertial measurement unit (IMU) sensor via ROS 2 (Robot Operating System), the system reliably detects the cones marking the track boundaries at distances of up to 35 m. Optimized trajectories are computed considering vehicle dynamics and simulated environmental factors such as visibility and lighting to navigate the track efficiently. The complete autonomous stack is implemented in ROS 2 and validated extensively in CARLA on a dedicated vehicle (ADS-DV) before being ported to the actual hardware, which includes the Jetson AGX Orin 64GB, ZED2i Stereo Camera, Robosense Helios 16P LiDAR, and CHCNAV Inertial Navigation System (INS).
☆ 6D Strawberry Pose Estimation: Real-time and Edge AI Solutions Using Purely Synthetic Training Data
Automated and selective harvesting of fruits has become an important area of research, particularly due to challenges such as high costs and a shortage of seasonal labor in advanced economies. This paper focuses on 6D pose estimation of strawberries using purely synthetic data generated through a procedural pipeline for photorealistic rendering. We employ the YOLOX-6D-Pose algorithm, a single-shot approach that leverages the YOLOX backbone, known for its balance between speed and accuracy, and its support for edge inference. To address the lacking availability of training data, we introduce a robust and flexible pipeline for generating synthetic strawberry data from various 3D models via a procedural Blender pipeline, where we focus on enhancing the realism of the synthesized data in comparison to previous work to make it a valuable resource for training pose estimation algorithms. Quantitative evaluations indicate that our models achieve comparable accuracy on both the NVIDIA RTX 3090 and Jetson Orin Nano across several ADD-S metrics, with the RTX 3090 demonstrating superior processing speed. However, the Jetson Orin Nano is particularly suited for resource-constrained environments, making it an excellent choice for deployment in agricultural robotics. Qualitative assessments further confirm the model's performance, demonstrating its capability to accurately infer the poses of ripe and partially ripe strawberries, while facing challenges in detecting unripe specimens. This suggests opportunities for future improvements, especially in enhancing detection capabilities for unripe strawberries (if desired) by exploring variations in color. Furthermore, the methodology presented could be adapted easily for other fruits such as apples, peaches, and plums, thereby expanding its applicability and impact in the field of agricultural automation.
☆ Experiences from Benchmarking Vision-Language-Action Models for Robotic Manipulation
Foundation models applied in robotics, particularly \textbf{Vision--Language--Action (VLA)} models, hold great promise for achieving general-purpose manipulation. Yet, systematic real-world evaluations and cross-model comparisons remain scarce. This paper reports our \textbf{empirical experiences} from benchmarking four representative VLAs -- \textbf{ACT}, \textbf{OpenVLA--OFT}, \textbf{RDT-1B}, and \boldmath{$π_0$} -- across four manipulation tasks conducted in both simulation and on the \textbf{ALOHA Mobile} platform. We establish a \textbf{standardized evaluation framework} that measures performance along three key dimensions: (1) \textit{accuracy and efficiency} (success rate and time-to-success), (2) \textit{adaptability} across in-distribution, spatial out-of-distribution, and instance-plus-spatial out-of-distribution settings, and (3) \textit{language instruction-following accuracy}. Through this process, we observe that \boldmath{$π_0$} demonstrates superior adaptability in out-of-distribution scenarios, while \textbf{ACT} provides the highest stability in-distribution. Further analysis highlights differences in computational demands, data-scaling behavior, and recurring failure modes such as near-miss grasps, premature releases, and long-horizon state drift. These findings reveal practical trade-offs among VLA model architectures in balancing precision, generalization, and deployment cost, offering actionable insights for selecting and deploying VLAs in real-world robotic manipulation tasks.
☆ Sashimi-Bot: Autonomous Tri-manual Advanced Manipulation and Cutting of Deformable Objects
Sverre Herland, Amit Parag, Elling Ruud Øye, Fangyi Zhang, Fouad Makiyeh, Aleksander Lillienskiold, Abhaya Pal Singh, Edward H. Adelson, Francois Chaumette, Alexandre Krupa, Peter Corke, Ekrem Misimi
Advanced robotic manipulation of deformable, volumetric objects remains one of the greatest challenges due to their pliancy, frailness, variability, and uncertainties during interaction. Motivated by these challenges, this article introduces Sashimi-Bot, an autonomous multi-robotic system for advanced manipulation and cutting, specifically the preparation of sashimi. The objects that we manipulate, salmon loins, are natural in origin and vary in size and shape, they are limp and deformable with poorly characterized elastoplastic parameters, while also being slippery and hard to hold. The three robots straighten the loin; grasp and hold the knife; cut with the knife in a slicing motion while cooperatively stabilizing the loin during cutting; and pick up the thin slices from the cutting board or knife blade. Our system combines deep reinforcement learning with in-hand tool shape manipulation, in-hand tool cutting, and feedback of visual and tactile information to achieve robustness to the variabilities inherent in this task. This work represents a milestone in robotic manipulation of deformable, volumetric objects that may inspire and enable a wide range of other real-world applications.
☆ Humanoid Whole-Body Badminton via Multi-Stage Reinforcement Learning
Humanoid robots have demonstrated strong capability for interacting with deterministic scenes across locomotion, manipulation, and more challenging loco-manipulation tasks. Yet the real world is dynamic, quasi-static interactions are insufficient to cope with the various environmental conditions. As a step toward more dynamic interaction scenario, we present a reinforcement-learning-based training pipeline that produces a unified whole-body controller for humanoid badminton, enabling coordinated lower-body footwork and upper-body striking without any motion priors or expert demonstrations. Training follows a three-stage curriculum: first footwork acquisition, then precision-guided racket swing generation, and finally task-focused refinement, yielding motions in which both legs and arms serve the hitting objective. For deployment, we incorporate an Extended Kalman Filter (EKF) to estimate and predict shuttlecock trajectories for target striking. We also introduce a prediction-free variant that dispenses with EKF and explicit trajectory prediction. To validate the framework, we conduct five sets of experiment in both simulation and the real world. In simulation, two robots sustain a rally of 21 consecutive hits. Moreover, the prediction-free variant achieves successful hits with comparable performance relative to the target-known policy. In real-world tests, both the prediction and controller module exhibit high accuracy, and on-court hitting achieves an outgoing shuttle speed up to 10 m/s with a mean return landing distance of 3.5 m. These experiment results show that our humanoid robot can deliver highly dynamic while precise goal striking in badminton, and can be adapted to more dynamism critical domains.
☆ Phys-Liquid: A Physics-Informed Dataset for Estimating 3D Geometry and Volume of Transparent Deformable Liquids AAAI-26
Estimating the geometric and volumetric properties of transparent deformable liquids is challenging due to optical complexities and dynamic surface deformations induced by container movements. Autonomous robots performing precise liquid manipulation tasks, such as dispensing, aspiration, and mixing, must handle containers in ways that inevitably induce these deformations, complicating accurate liquid state assessment. Current datasets lack comprehensive physics-informed simulation data representing realistic liquid behaviors under diverse dynamic scenarios. To bridge this gap, we introduce Phys-Liquid, a physics-informed dataset comprising 97,200 simulation images and corresponding 3D meshes, capturing liquid dynamics across multiple laboratory scenes, lighting conditions, liquid colors, and container rotations. To validate the realism and effectiveness of Phys-Liquid, we propose a four-stage reconstruction and estimation pipeline involving liquid segmentation, multi-view mask generation, 3D mesh reconstruction, and real-world scaling. Experimental results demonstrate improved accuracy and consistency in reconstructing liquid geometry and volume, outperforming existing benchmarks. The dataset and associated validation methods facilitate future advancements in transparent liquid perception tasks. The dataset and code are available at https://dualtransparency.github.io/Phys-Liquid/.
comment: 14 pages, 19 figures. Accepted as an oral paper at AAAI-26 (Main Technical Track). Code and dataset: https://github.com/dualtransparency/Phys-Liquid-AAAI Project page: https://dualtransparency.github.io/Phys-Liquid/
☆ AdaptPNP: Integrating Prehensile and Non-Prehensile Skills for Adaptive Robotic Manipulation
Jinxuan Zhu, Chenrui Tie, Xinyi Cao, Yuran Wang, Jingxiang Guo, Zixuan Chen, Haonan Chen, Junting Chen, Yangyu Xiao, Ruihai Wu, Lin Shao
Non-prehensile (NP) manipulation, in which robots alter object states without forming stable grasps (for example, pushing, poking, or sliding), significantly broadens robotic manipulation capabilities when grasping is infeasible or insufficient. However, enabling a unified framework that generalizes across different tasks, objects, and environments while seamlessly integrating non-prehensile and prehensile (P) actions remains challenging: robots must determine when to invoke NP skills, select the appropriate primitive for each context, and compose P and NP strategies into robust, multi-step plans. We introduce ApaptPNP, a vision-language model (VLM)-empowered task and motion planning framework that systematically selects and combines P and NP skills to accomplish diverse manipulation objectives. Our approach leverages a VLM to interpret visual scene observations and textual task descriptions, generating a high-level plan skeleton that prescribes the sequence and coordination of P and NP actions. A digital-twin based object-centric intermediate layer predicts desired object poses, enabling proactive mental rehearsal of manipulation sequences. Finally, a control module synthesizes low-level robot commands, with continuous execution feedback enabling online task plan refinement and adaptive replanning through the VLM. We evaluate ApaptPNP across representative P&NP hybrid manipulation tasks in both simulation and real-world environments. These results underscore the potential of hybrid P&NP manipulation as a crucial step toward general-purpose, human-level robotic manipulation capabilities. Project Website: https://sites.google.com/view/adaptpnp/home
☆ Autonomous Vehicle Path Planning by Searching With Differentiable Simulation
Planning allows an agent to safely refine its actions before executing them in the real world. In autonomous driving, this is crucial to avoid collisions and navigate in complex, dense traffic scenarios. One way to plan is to search for the best action sequence. However, this is challenging when all necessary components - policy, next-state predictor, and critic - have to be learned. Here we propose Differentiable Simulation for Search (DSS), a framework that leverages the differentiable simulator Waymax as both a next state predictor and a critic. It relies on the simulator's hardcoded dynamics, making state predictions highly accurate, while utilizing the simulator's differentiability to effectively search across action sequences. Our DSS agent optimizes its actions using gradient descent over imagined future trajectories. We show experimentally that DSS - the combination of planning gradients and stochastic search - significantly improves tracking and path planning accuracy compared to sequence prediction, imitation learning, model-free RL, and other planning methods.
☆ Miniature Testbed for Validating Multi-Agent Cooperative Autonomous Driving
Hyunchul Bae, Eunjae Lee, Jehyeop Han, Minhee Kang, Jaehyeon Kim, Junggeun Seo, Minkyun Noh, Heejin Ahn
Cooperative autonomous driving, which extends vehicle autonomy by enabling real-time collaboration between vehicles and smart roadside infrastructure, remains a challenging yet essential problem. However, none of the existing testbeds employ smart infrastructure equipped with sensing, edge computing, and communication capabilities. To address this gap, we design and implement a 1:15-scale miniature testbed, CIVAT, for validating cooperative autonomous driving, consisting of a scaled urban map, autonomous vehicles with onboard sensors, and smart infrastructure. The proposed testbed integrates V2V and V2I communication with the publish-subscribe pattern through a shared Wi-Fi and ROS2 framework, enabling information exchange between vehicles and infrastructure to realize cooperative driving functionality. As a case study, we validate the system through infrastructure-based perception and intersection management experiments.
comment: 8 pages
☆ Latent-Space Autoregressive World Model for Efficient and Robust Image-Goal Navigation
Traditional navigation methods rely heavily on accurate localization and mapping. In contrast, world models that capture environmental dynamics in latent space have opened up new perspectives for navigation tasks, enabling systems to move beyond traditional multi-module pipelines. However, world model often suffers from high computational costs in both training and inference. To address this, we propose LS-NWM - a lightweight latent space navigation world model that is trained and operates entirely in latent space, compared to the state-of-the-art baseline, our method reduces training time by approximately 3.2x and planning time by about 447x,while further improving navigation performance with a 35% higher SR and an 11% higher SPL. The key idea is that accurate pixel-wise environmental prediction is unnecessary for navigation. Instead, the model predicts future latent states based on current observational features and action inputs, then performs path planning and decision-making within this compact representation, significantly improving computational efficiency. By incorporating an autoregressive multi-frame prediction strategy during training, the model effectively captures long-term spatiotemporal dependencies, thereby enhancing navigation performance in complex scenarios. Experimental results demonstrate that our method achieves state-of-the-art navigation performance while maintaining a substantial efficiency advantage over existing approaches.
☆ Dynamic Reconfiguration of Robotic Swarms: Coordination and Control for Precise Shape Formation
Coordination of movement and configuration in robotic swarms is a challenging endeavor. Deciding when and where each individual robot must move is a computationally complex problem. The challenge is further exacerbated by difficulties inherent to physical systems, such as measurement error and control dynamics. Thus, how to best determine the optimal path for each robot, when moving from one configuration to another, and how to best perform such determination and effect corresponding motion remains an open problem. In this paper, we show an algorithm for such coordination of robotic swarms. Our methods allow seamless transition from one configuration to another, leveraging geometric formulations that are mapped to the physical domain through appropriate control, localization, and mapping techniques. This paves the way for novel applications of robotic swarms by enabling more sophisticated distributed behaviors.
comment: accepted at the 9th International Conference on Algorithms, Computing and Systems (ICACS 2025)
☆ Dexterous Manipulation Transfer via Progressive Kinematic-Dynamic Alignment AAAI 2026
The inherent difficulty and limited scalability of collecting manipulation data using multi-fingered robot hand hardware platforms have resulted in severe data scarcity, impeding research on data-driven dexterous manipulation policy learning. To address this challenge, we present a hand-agnostic manipulation transfer system. It efficiently converts human hand manipulation sequences from demonstration videos into high-quality dexterous manipulation trajectories without requirements of massive training data. To tackle the multi-dimensional disparities between human hands and dexterous hands, as well as the challenges posed by high-degree-of-freedom coordinated control of dexterous hands, we design a progressive transfer framework: first, we establish primary control signals for dexterous hands based on kinematic matching; subsequently, we train residual policies with action space rescaling and thumb-guided initialization to dynamically optimize contact interactions under unified rewards; finally, we compute wrist control trajectories with the objective of preserving operational semantics. Using only human hand manipulation videos, our system automatically configures system parameters for different tasks, balancing kinematic matching and dynamic optimization across dexterous hands, object categories, and tasks. Extensive experimental results demonstrate that our framework can automatically generate smooth and semantically correct dexterous hand manipulation that faithfully reproduces human intentions, achieving high efficiency and strong generalizability with an average transfer success rate of 73%, providing an easily implementable and scalable method for collecting robot dexterous manipulation data.
comment: 13 pages, 15 figures. Accepted by AAAI 2026
☆ Terradynamics and design of tip-extending robotic anchors
Deniz Kerimoglu, Nicholas D. Naclerio, Sean Chu, Andrew Krohn, Vineet Kupunaram, Alexander Schepelmann, Daniel I. Goldman, Elliot W. Hawkes
Most engineered pilings require substantially more force to be driven into the ground than they can resist during extraction. This requires relatively heavy equipment for insertion, which is problematic for anchoring in hard-to-access sites, including in extraterrestrial locations. In contrast, for tree roots, the external reaction force required to extract is much greater than required to insert--little more than the weight of the seed initiates insertion. This is partly due to the mechanism by which roots insert into the ground: tip extension. Proof-of-concept robotic prototypes have shown the benefits of using this mechanism, but a rigorous understanding of the underlying granular mechanics and how they inform the design of a robotic anchor is lacking. Here, we study the terradynamics of tip-extending anchors compared to traditional piling-like intruders, develop a set of design insights, and apply these to create a deployable robotic anchor. Specifically, we identify that to increase an anchor's ratio of extraction force to insertion force, it should: (i) extend beyond a critical depth; (ii) include hair-like protrusions; (iii) extend near-vertically, and (iv) incorporate multiple smaller anchors rather than a single large anchor. Synthesizing these insights, we developed a lightweight, soft robotic, root-inspired anchoring device that inserts into the ground with a reaction force less than its weight. We demonstrate that the 300 g device can deploy a series of temperature sensors 45 cm deep into loose Martian regolith simulant while anchoring with an average of 120 N, resulting in an anchoring-to-weight ratio of 40:1.
☆ Collaborative Multi-Robot Non-Prehensile Manipulation via Flow-Matching Co-Generation
Coordinating a team of robots to reposition multiple objects in cluttered environments requires reasoning jointly about where robots should establish contact, how to manipulate objects once contact is made, and how to navigate safely and efficiently at scale. Prior approaches typically fall into two extremes -- either learning the entire task or relying on privileged information and hand-designed planners -- both of which struggle to handle diverse objects in long-horizon tasks. To address these challenges, we present a unified framework for collaborative multi-robot, multi-object non-prehensile manipulation that integrates flow-matching co-generation with anonymous multi-robot motion planning. Within this framework, a generative model co-generates contact formations and manipulation trajectories from visual observations, while a novel motion planner conveys robots at scale. Crucially, the same planner also supports coordination at the object level, assigning manipulated objects to larger target structures and thereby unifying robot- and object-level reasoning within a single algorithmic framework. Experiments in challenging simulated environments demonstrate that our approach outperforms baselines in both motion planning and manipulation tasks, highlighting the benefits of generative co-design and integrated planning for scaling collaborative manipulation to complex multi-agent, multi-object settings. Visit gco-paper.github.io for code and demonstrations.
☆ WetExplorer: Automating Wetland Greenhouse-Gas Surveys with an Autonomous Mobile Robot
Quantifying greenhouse-gases (GHG) in wetlands is critical for climate modeling and restoration assessment, yet manual sampling is labor-intensive, and time demanding. We present WetExplorer, an autonomous tracked robot that automates the full GHG-sampling workflow. The robot system integrates low-ground-pressure locomotion, centimeter-accurate lift placement, dual-RTK sensor fusion, obstacle avoidance planning, and deep-learning perception in a containerized ROS2 stack. Outdoor trials verified that the sensor-fusion stack maintains a mean localization error of 1.71 cm, the vision module estimates object pose with 7 mm translational and 3° rotational accuracy, while indoor trials demonstrated that the full motion-planning pipeline positions the sampling chamber within a global tolerance of 70 mm while avoiding obstacles, all without human intervention. By eliminating the manual bottleneck, WetExplorer enables high-frequency, multi-site GHG measurements and opens the door for dense, long-duration datasets in saturated wetland terrain.
comment: To be published in 2025 IEEE International Conference on Robotics and Biomimetics
♻ ☆ EAST: Environment Aware Safe Tracking using Planning and Control Co-Design
This paper considers the problem of autonomous mobile robot navigation in unknown environments with moving obstacles. We propose a new method to achieve environment-aware safe tracking (EAST) of robot motion plans that integrates an obstacle clearance cost for path planning, a convex reachable set for robot motion prediction, and safety constraints for dynamic obstacle avoidance. EAST adapts the motion of the robot according to the locally sensed environment geometry and dynamics, leading to fast motion in wide open areas and cautious behavior in narrow passages or near moving obstacles. Our control design uses a reference governor, a virtual dynamical system that guides the robot's motion and decouples the path tracking and safety objectives. While reference governor methods have been used for safe tracking control in static environments, our key contribution is an extension to dynamic environments using convex optimization with control barrier function (CBF) constraints. Thus, our work establishes a connection between reference governor techniques and CBF techniques for safe control in dynamic environments. We validate our approach in simulated and real-world environments, featuring complex obstacle configurations and natural dynamic obstacle motion.
♻ ☆ Sensory-Motor Control with Large Language Models via Iterative Policy Refinement
We propose a method that enables large language models (LLMs) to control embodied agents through the generation of control policies that directly map continuous observation vectors to continuous action vectors. At the outset, the LLMs generate a control strategy based on a textual description of the agent, its environment, and the intended goal. This strategy is then iteratively refined through a learning process in which the LLMs are repeatedly prompted to improve the current strategy, using performance feedback and sensory-motor data collected during its evaluation. The method is validated on classic control tasks from the Gymnasium library and the inverted pendulum task from the MuJoCo library. The approach proves effective with relatively compact models such as GPT-oss:120b and Qwen2.5:72b. In most cases, it successfully identifies optimal or near-optimal solutions by integrating symbolic knowledge derived through reasoning with sub-symbolic sensory-motor data gathered as the agent interacts with its environment.
comment: Article updated with results from gpt-oss:120b and gpt-oss:20b. 27 pages (13 pages are from appendix), 8 figures, 2 tables, code for experiments replication and supplementary material provided at https://github.com/jtyska/llm-robotics-article/
♻ ☆ DiAReL: Reinforcement Learning with Disturbance Awareness for Robust Sim2Real Policy Transfer in Robot Control
Delayed Markov decision processes (DMDPs) fulfill the Markov property by augmenting the state space of agents with a finite time window of recently committed actions. In reliance on these state augmentations, delay-resolved reinforcement learning algorithms train policies to learn optimal interactions with environments featuring observation or action delays. Although such methods can be directly trained on the real robots, due to sample inefficiency, limited resources, or safety constraints, a common approach is to transfer models trained in simulation to the physical robot. However, robotic simulations rely on approximated models of the physical systems, which hinders the sim2real transfer. In this work, we consider various uncertainties in modeling the robot or environment dynamics as unknown intrinsic disturbances applied to the system input. We introduce the disturbance-augmented Markov decision process (DAMDP) in delayed settings as a novel representation to incorporate disturbance estimation in training on-policy reinforcement learning algorithms. The proposed method is validated across several metrics on learning robotic reaching and pushing tasks and compared with disturbance-unaware baselines. The results show that the disturbance-augmented models can achieve higher stabilization and robustness in the control response, which in turn improves the prospects of successful sim2real transfer.
comment: Accepted for publication in IEEE Transactions on Control Systems Technology (TCST)
♻ ☆ Dynamic Sparsity: Challenging Common Sparsity Assumptions for Learning World Models in Robotic Reinforcement Learning Benchmarks
Muthukumar Pandaram, Jakob Hollenstein, David Drexel, Samuele Tosatto, Antonio Rodríguez-Sánchez, Justus Piater
The use of learned dynamics models, also known as world models, can improve the sample efficiency of reinforcement learning. Recent work suggests that the underlying causal graphs of such dynamics models are sparsely connected, with each of the future state variables depending only on a small subset of the current state variables, and that learning may therefore benefit from sparsity priors. Similarly, temporal sparsity, i.e. sparsely and abruptly changing local dynamics, has also been proposed as a useful inductive bias.
In this work, we critically examine these assumptions by analyzing ground-truth dynamics from a set of robotic reinforcement learning environments in the MuJoCo Playground benchmark suite, aiming to determine whether the proposed notions of state and temporal sparsity actually tend to hold in typical reinforcement learning tasks.
We study (i) whether the causal graphs of environment dynamics are sparse, (ii) whether such sparsity is state-dependent, and (iii) whether local system dynamics change sparsely.
Our results indicate that global sparsity is rare, but instead the tasks show local, state-dependent sparsity in their dynamics and this sparsity exhibits distinct structures, appearing in temporally localized clusters (e.g., during contact events) and affecting specific subsets of state dimensions. These findings challenge common sparsity prior assumptions in dynamics learning, emphasizing the need for grounded inductive biases that reflect the state-dependent sparsity structure of real-world dynamics.
♻ ☆ Reconfigurable hydrostatics: Toward versatile and efficient load-bearing robotics
Wearable and legged robot designers face multiple challenges when choosing actuation. Traditional fully actuated designs using electric motors are multifunctional but oversized and inefficient for bearing conservative loads and for being backdrivable. Alternatively, quasi-passive and underactuated designs reduce the amount of motorization and energy storage, but are often designed for specific tasks. Designers of versatile and stronger wearable robots will face these challenges unless future actuators become very torque-dense, backdrivable and efficient
This paper explores a design paradigm for addressing this issue: reconfigurable hydrostatics. We show that a hydrostatic actuator can integrate a passive force mechanism and a sharing mechanism in the fluid domain and still be multifunctional. First, an analytical study compares the effect of these two mechanisms on the motorization requirements in the context of a load-bearing exoskeleton. Then, the hydrostatic concept integrating these two mechanisms using hydraulic components is presented. A case study analysis shows the mass/efficiency/inertia benefits of the concept over a fully actuated one. Then, experiments are conducted on robotic legs to demonstrate that the actuator concept can meet the expected performance in terms of force tracking, versatility, and efficiency under controlled conditions. The proof-of-concept can track the vertical ground reaction force (GRF) profiles of walking, running, squatting, and jumping, and the energy consumption is 4.8x lower for walking. The transient force behaviors due to switching from one leg to the other are also analyzed along with some mitigation to improve them.
♻ ☆ Enhancing the NAO: Extending Capabilities of Legacy Robots for Long-Term Research
Legacy (unsupported) robotic platforms often lose research utility when manufacturer support ends, preventing integration of modern sensing, speech, and interaction capabilities. We present the Enhanced NAO, a revitalized version of Aldebaran's NAO robot featuring upgraded beamforming microphones, RGB-D and thermal cameras, and additional compute resources in a fully self-contained package. This system combines cloud-based and local models for perception and dialogue, while preserving the NAO's expressive body and behaviors. In a pilot user study validating conversational performance, the Enhanced NAO delivered significantly higher conversational quality and elicited stronger user preference compared to the NAO AI Edition, without increasing response latency. The added visual and thermal sensing modalities established a foundation for future perception-driven interaction. Beyond this implementation, our framework provides a platform-agnostic strategy for extending the lifespan and research utility of legacy robots, ensuring they remain valuable tools for human-robot interaction.
♻ ☆ Large Language Model-assisted Autonomous Vehicle Recovery from Immobilization
Despite significant advancements in recent decades, autonomous vehicles (AVs) continue to face challenges in navigating certain traffic scenarios where human drivers excel. In such situations, AVs often become immobilized, disrupting overall traffic flow. Current recovery solutions, such as remote intervention (which is costly and inefficient) and manual takeover (which excludes non-drivers and limits AV accessibility), are inadequate. This paper introduces StuckSolver, a novel Large Language Model (LLM) driven recovery framework that enables AVs to resolve immobilization scenarios through self-reasoning and/or passenger-guided decision-making. StuckSolver is designed as a plug-in add-on module that operates on top of the AV's existing perception-planning-control stack, requiring no modification to its internal architecture. Instead, it interfaces with standard sensor data streams to detect immobilization states, interpret environmental context, and generate high-level recovery commands that can be executed by the AV's native planner. We evaluate StuckSolver on the Bench2Drive benchmark and in custom-designed uncertainty scenarios. Results show that StuckSolver achieves near-state-of-the-art performance through autonomous self-reasoning alone and exhibits further improvements when passenger guidance is incorporated.
comment: 7 pages
♻ ☆ A Learning-Based Framework for Collision-Free Motion Planning
This paper presents a learning-based extension to a Circular Field (CF)-based motion planner for efficient, collision-free trajectory generation in cluttered environments. The proposed approach overcomes the limitations of hand-tuned force field parameters by employing a deep neural network trained to infer optimal planner gains from a single depth image of the scene. The pipeline incorporates a CUDA-accelerated perception module, a predictive agent-based planning strategy, and a dataset generated through Bayesian optimization in simulation. The resulting framework enables real-time planning without manual parameter tuning and is validated both in simulation and on a Franka Emika Panda robot. Experimental results demonstrate successful task completion and improved generalization compared to classical planners.
♻ ☆ Pelican-VL 1.0: A Foundation Brain Model for Embodied Intelligence
Yi Zhang, Che Liu, Xiancong Ren, Hanchu Ni, Shuai Zhang, Zeyuan Ding, Jiayu Hu, Hanzhe Shan, Zhenwei Niu, Zhaoyang Liu, Shuang Liu, Yue Zhao, Junbo Qi, Qinfan Zhang, Dengjie Li, Yidong Wang, Jiachen Luo, Yong Dai, Zenglin Xu, Bin Shen, Qifan Wang, Jian Tang, Xiaozhu Ju
This report presents Pelican-VL 1.0, a new family of open-source embodied brain models with parameter scales ranging from 7 billion to 72 billion. Our explicit mission is clearly stated as: To embed powerful intelligence into various embodiments. Pelican-VL 1.0 is currently the largest-scale open-source embodied multimodal brain model. Its core advantage lies in the in-depth integration of data power and intelligent adaptive learning mechanisms. Specifically, metaloop distilled a high-quality dataset from a raw dataset containing 4+ billion tokens. Pelican-VL 1.0 is trained on a large-scale cluster of 1000+ A800 GPUs, consuming over 50k+ A800 GPU-hours per checkpoint. This translates to a 20.3% performance uplift from its base model and outperforms 100B-level open-source counterparts by 10.6%, placing it on par with leading proprietary systems on well-known embodied benchmarks. We establish a novel framework, DPPO (Deliberate Practice Policy Optimization), inspired by human metacognition to train Pelican-VL 1.0. We operationalize this as a metaloop that teaches the AI to practice deliberately, which is a RL-Refine-Diagnose-SFT loop.
♻ ☆ Leveraging Sidewalk Robots for Walkability-Related Analyses
Walkability is a key component of sustainable urban development. In walkability studies, collecting detailed pedestrian infrastructure data remains challenging due to the high costs and limited scalability of traditional methods. Sidewalk delivery robots, increasingly deployed in urban environments, offer a promising solution to these limitations. This paper explores how these robots can serve as mobile data collection platforms, capturing sidewalk-level features related to walkability in a scalable, automated, and real-time manner. A sensor-equipped robot was deployed on a sidewalk network at KTH in Stockholm, completing 101 trips covering 900 segment records. From the collected data, different typologies of features are derived, including robot trip characteristics (e.g., speed, duration), sidewalk conditions (e.g., width, surface unevenness), and sidewalk utilization (e.g., pedestrian density). Their walkability-related implications were investigated with a series of analyses. The results demonstrate that pedestrian movement patterns are strongly influenced by sidewalk characteristics, with higher density, reduced width, and surface irregularity associated with slower and more variable trajectories. Notably, robot speed closely mirrors pedestrian behavior, highlighting its potential as a proxy for assessing pedestrian dynamics. The proposed framework enables continuous monitoring of sidewalk conditions and pedestrian behavior, contributing to the development of more walkable, inclusive, and responsive urban environments.
♻ ☆ TTF-VLA: Temporal Token Fusion via Pixel-Attention Integration for Vision-Language-Action Models AAAI 2026
Vision-Language-Action (VLA) models process visual inputs independently at each timestep, discarding valuable temporal information inherent in robotic manipulation tasks. This frame-by-frame processing makes models vulnerable to visual noise while ignoring the substantial coherence between consecutive frames in manipulation sequences. We propose Temporal Token Fusion (TTF), a training-free approach that intelligently integrates historical and current visual representations to enhance VLA inference quality. Our method employs dual-dimension detection combining efficient grayscale pixel difference analysis with attention-based semantic relevance assessment, enabling selective temporal token fusion through hard fusion strategies and keyframe anchoring to prevent error accumulation. Comprehensive experiments across LIBERO, SimplerEnv, and real robot tasks demonstrate consistent improvements: 4.0 percentage points average on LIBERO (72.4\% vs 68.4\% baseline), cross-environment validation on SimplerEnv (4.8\% relative improvement), and 8.7\% relative improvement on real robot tasks. Our approach proves model-agnostic, working across OpenVLA and VLA-Cache architectures. Notably, TTF reveals that selective Query matrix reuse in attention mechanisms enhances rather than compromises performance, suggesting promising directions for direct KQV matrix reuse strategies that achieve computational acceleration while improving task success rates.
comment: Accepted to AAAI 2026. Camera-ready version
♻ ☆ MSGNav: Unleashing the Power of Multi-modal 3D Scene Graph for Zero-Shot Embodied Navigation
Xun Huang, Shijia Zhao, Yunxiang Wang, Xin Lu, Wanfa Zhang, Rongsheng Qu, Weixin Li, Yunhong Wang, Chenglu Wen
Embodied navigation is a fundamental capability for robotic agents operating. Real-world deployment requires open vocabulary generalization and low training overhead, motivating zero-shot methods rather than task-specific RL training. However, existing zero-shot methods that build explicit 3D scene graphs often compress rich visual observations into text-only relations, leading to high construction cost, irreversible loss of visual evidence, and constrained vocabularies. To address these limitations, we introduce the Multi-modal 3D Scene Graph (M3DSG), which preserves visual cues by replacing textual relation
comment: 10 pages
♻ ☆ Towards Efficient Certification of Maritime Remote Operation Centers
Christian Neurohr, Marcel Saager, Lina Putze, Jan-Patrick Osterloh, Karina Rothemann, Hilko Wiards, Eckard Böde, Axel Hahn
Additional automation being build into ships implies a shift of crew from ship to shore. However, automated ships still have to be monitored and, in some situations, controlled remotely. These tasks are carried out by human operators located in shore-based remote operation centers. In this work, we present a concept for a hazard database that supports the safeguarding and certification of such remote operation centers. The concept is based on a categorization of hazard sources which we derive from a generic functional architecture. A subsequent preliminary suitability analysis unveils which methods for hazard analysis and risk assessment can adequately fill this hazard database.
♻ ☆ Efficient Learning-Based Control of a Legged Robot in Lunar Gravity
Legged robots are promising candidates for exploring challenging areas on low-gravity bodies such as the Moon, Mars, or asteroids, thanks to their advanced mobility on unstructured terrain. However, as planetary robots' power and thermal budgets are highly restricted, these robots need energy-efficient control approaches that easily transfer to multiple gravity environments. In this work, we introduce a reinforcement learning-based control approach for legged robots with gravity-scaled power-optimized reward functions. We use our approach to develop and validate a locomotion controller and a base pose controller in gravity environments from lunar gravity (1.62 m/s2) to a hypothetical super-Earth (19.62 m/s2). Our approach successfully scales across these gravity levels for locomotion and base pose control with the gravity-scaled reward functions. The power-optimized locomotion controller reached a power consumption for locomotion of 23.4 W in Earth gravity on a 15.65 kg robot at 0.4 m/s, a 23 % improvement over the baseline policy. Additionally, we designed a constant-force spring offload system that allowed us to conduct real-world experiments on legged locomotion in lunar gravity. In lunar gravity, the power-optimized control policy reached 12.2 W, 36 % less than a baseline controller which is not optimized for power efficiency. Our method provides a scalable approach to developing power-efficient locomotion controllers for legged robots across multiple gravity levels.
♻ ☆ Multistep Quasimetric Learning for Scalable Goal-conditioned Reinforcement Learning
Learning how to reach goals in an environment is a longstanding challenge in AI, yet reasoning over long horizons remains a challenge for modern methods. The key question is how to estimate the temporal distance between pairs of observations. While temporal difference methods leverage local updates to provide optimality guarantees, they often perform worse than Monte Carlo methods that perform global updates (e.g., with multi-step returns), which lack such guarantees. We show how these approaches can be integrated into a practical GCRL method that fits a quasimetric distance using a multistep Monte-Carlo return. We show our method outperforms existing GCRL methods on long-horizon simulated tasks with up to 4000 steps, even with visual observations. We also demonstrate that our method can enable stitching in the real-world robotic manipulation domain (Bridge setup). Our approach is the first end-to-end GCRL method that enables multistep stitching in this real-world manipulation domain from an unlabeled offline dataset of visual observations.
♻ ☆ Zero-Shot Temporal Interaction Localization for Egocentric Videos IROS 2025
Locating human-object interaction (HOI) actions within video serves as the foundation for multiple downstream tasks, such as human behavior analysis and human-robot skill transfer. Current temporal action localization methods typically rely on annotated action and object categories of interactions for optimization, which leads to domain bias and low deployment efficiency. Although some recent works have achieved zero-shot temporal action localization (ZS-TAL) with large vision-language models (VLMs), their coarse-grained estimations and open-loop pipelines hinder further performance improvements for temporal interaction localization (TIL). To address these issues, we propose a novel zero-shot TIL approach dubbed EgoLoc to locate the timings of grasp actions for human-object interaction in egocentric videos. EgoLoc introduces a self-adaptive sampling strategy to generate reasonable visual prompts for VLM reasoning. By absorbing both 2D and 3D observations, it directly samples high-quality initial guesses around the possible contact/separation timestamps of HOI according to 3D hand velocities, leading to high inference accuracy and efficiency. In addition, EgoLoc generates closed-loop feedback from visual and dynamic cues to further refine the localization results. Comprehensive experiments on the publicly available dataset and our newly proposed benchmark demonstrate that EgoLoc achieves better temporal interaction localization for egocentric videos compared to state-of-the-art baselines. We have released our code and relevant data as open-source at https://github.com/IRMVLab/EgoLoc.
comment: Accepted to IROS 2025
♻ ☆ Fast Finite-Time Sliding Mode Control for Chattering-Free Trajectory Tracking of Robotic Manipulators
Achieving precise and efficient trajectory tracking in robotic arms remains a key challenge due to system uncertainties and chattering effects in conventional sliding mode control (SMC). This paper presents a chattering-free fast terminal sliding mode control (FTSMC) strategy for a three-degree-of-freedom (3-DOF) robotic arm, designed to enhance tracking accuracy and robustness while ensuring finite-time convergence. The control framework is developed using Newton-Euler dynamics, followed by a state-space representation that captures the system's angular position and velocity. By incorporating an improved sliding surface and a Lyapunov-based stability analysis, the proposed FTSMC effectively mitigates chattering while preserving the advantages of SMC, such as fast response and strong disturbance rejection. The controller's performance is rigorously evaluated through comparisons with conventional PD sliding mode control (PDSMC) and terminal sliding mode control (TSMC). Simulation results demonstrate that the proposed approach achieves superior trajectory tracking performance, faster convergence, and enhanced stability compared to existing methods, making it a promising solution for high-precision robotic applications.
♻ ☆ Convergent Functions, Divergent Forms
We introduce LOKI, a compute-efficient framework for co-designing morphologies and control policies that generalize across unseen tasks. Inspired by biological adaptation -- where animals quickly adjust to morphological changes -- our method overcomes the inefficiencies of traditional evolutionary and quality-diversity algorithms. We propose learning convergent functions: shared control policies trained across clusters of morphologically similar designs in a learned latent space, drastically reducing the training cost per design. Simultaneously, we promote divergent forms by replacing mutation with dynamic local search, enabling broader exploration and preventing premature convergence. The policy reuse allows us to explore 780$\times$ more designs using 78% fewer simulation steps and 40% less compute per design. Local competition paired with a broader search results in a diverse set of high-performing final morphologies. Using the UNIMAL design space and a flat-terrain locomotion task, LOKI discovers a rich variety of designs -- ranging from quadrupeds to crabs, bipedals, and spinners -- far more diverse than those produced by prior work. These morphologies also transfer better to unseen downstream tasks in agility, stability, and manipulation domains (e.g., 2$\times$ higher reward on bump and push box incline tasks). Overall, our approach produces designs that are both diverse and adaptable, with substantially greater sample efficiency than existing co-design methods. (Project website: https://loki-codesign.github.io/)
♻ ☆ Harnessing Bounded-Support Evolution Strategies for Policy Refinement
Improving competent robot policies with on-policy RL is often hampered by noisy, low-signal gradients. We revisit Evolution Strategies (ES) as a policy-gradient proxy and localize exploration with bounded, antithetic triangular perturbations, suitable for policy refinement. We propose Triangular-Distribution ES (TD-ES) which pairs bounded triangular noise with a centered-rank finite-difference estimator to deliver stable, parallelizable, gradient-free updates. In a two-stage pipeline - PPO pretraining followed by TD-ES refinement - this preserves early sample efficiency while enabling robust late-stage gains. Across a suite of robotic manipulation tasks, TD-ES raises success rates by 26.5% relative to PPO and greatly reduces variance, offering a simple, compute-light path to reliable refinement.
comment: 10 pages, 6 figures, to be published in Australasian Conference on Robotics and Automation (ACRA 2025)
♻ ☆ Your Ride, Your Rules: Psychology and Cognition Enabled Automated Driving Systems
Despite rapid advances in autonomous driving technology, current autonomous vehicles (AVs) lack effective bidirectional human-machine communication, limiting their ability to personalize the riding experience and recover from uncertain or immobilized states. This limitation undermines occupant comfort and trust, potentially hindering the adoption of AV technologies. We propose PACE-ADS (Psychology and Cognition Enabled Automated Driving Systems), a human-centered autonomy framework enabling AVs to sense, interpret, and respond to both external traffic conditions and internal occupant states. PACE-ADS uses an agentic workflow where three foundation model agents collaborate: the Driver Agent interprets the external environment; the Psychologist Agent decodes passive psychological signals (e.g., EEG, heart rate, facial expressions) and active cognitive inputs (e.g., verbal commands); and the Coordinator Agent synthesizes these inputs to generate high-level decisions that enhance responsiveness and personalize the ride. PACE-ADS complements, rather than replaces, conventional AV modules. It operates at the semantic planning layer, while delegating low-level control to native systems. The framework activates only when changes in the rider's psychological state are detected or when occupant instructions are issued. It integrates into existing AV platforms with minimal adjustments, positioning PACE-ADS as a scalable enhancement. We evaluate it in closed-loop simulations across diverse traffic scenarios, including intersections, pedestrian interactions, work zones, and car-following. Results show improved ride comfort, dynamic behavioral adjustment, and safe recovery from edge-case scenarios via autonomous reasoning or rider input. PACE-ADS bridges the gap between technical autonomy and human-centered mobility.
comment: 32 pages, one colummns