Nikolaus Correll

Sarah Aguasvivas Manzano, Vani Sundaram, Artemis Xu et al.

We present formulation and open-source tools to achieve in-material model predictive control of sensor/actuator systems using learned forward kinematics and on-device computation. Microcontroller units (MCUs) that compute the prediction and control task while colocated with the sensors and actuators enable in-material untethered behaviors. In this approach, small parameter size neural network models learn forward kinematics offline. Our open-source compiler, nn4mc, generates code to offload these predictions onto MCUs. A Newton-Raphson solver then computes the control input in real time. We first benchmark this nonlinear control approach against a PID controller on a mass-spring-damper simulation. We then study experimental results on two experimental rigs with different sensing, actuation and computational hardware: a tendon-based platform with embedded LightLace sensors and a HASEL-based platform with magnetic sensors. Experimental results indicate effective high-bandwidth tracking of reference paths (greater than or equal to 120 Hz) with a small memory footprint (less than or equal to 6.4% of flash memory). The measured path following error does not exceed 2mm in the tendon-based platform. The simulated path following error does not exceed 1mm in the HASEL-based platform. The mean power consumption of this approach in an ARM Cortex-M4f device is 45.4 mW. This control approach is also compatible with Tensorflow Lite models and equivalent on-device code. In-material intelligence enables a new class of composites that infuse autonomy into structures and systems with refined artificial proprioception.

Zachary Allen, Max Conway, Lyle Antieau et al.

Electric vehicles (EV) create an urgent need for scalable battery recycling, yet disassembly of EV battery packs remains largely manual due to high design variability. We present our Robotic Agentic Platform for Intelligent Disassembly (RAPID), designed to investigate perception-driven manipulation, flexible automation, and AI-assisted robot programming in realistic recycling scenarios. The system integrates a gantry-mounted industrial manipulator, RGB-D perception, and an automated nut-running tool for fastener removal on a full-scale EV battery pack. An open-vocabulary object detection pipeline achieves 0.9757 mAP50, enabling reliable identification of screws, nuts, busbars, and other components. We experimentally evaluate (n=204) three one-shot fastener removal strategies: taught-in poses (97% success rate, 24 min duration), one-shot vision execution (57%, 29 min), and visual servoing (83%, 36 min), comparing success rate and disassembly time for the battery's top cover fasteners. To support flexible interaction, we introduce agentic AI specifications for robotic disassembly tasks, allowing LLM agents to translate high-level instructions into robot actions through structured tool interfaces and ROS services. We evaluate SmolAgents with GPT-4o-mini and Qwen 3.5 9B/4B on edge hardware. Tool-based interfaces achieve 100% task completion, while automatic ROS service discovery shows 43.3% failure rates, highlighting the need for structured robot APIs for reliable LLM-driven control. This open-source platform enables systematic investigation of human-robot collaboration, agentic robot programming, and increasingly autonomous disassembly workflows, providing a practical foundation for research toward scalable robotic battery recycling.

Artemis Shaw, Chen Liu, Justin Costa et al.

We present a bimanual mobile manipulator built on the open-source XLeRobot with integrated onboard compute for less than \$1300. Key contributions include: (1) optimized mechanical design maximizing stiffness-to-weight ratio, (2) a Tri-Bus power topology isolating compute from motor-induced voltage transients, and (3) embedded autonomy using NVIDIA Jetson Orin Nano for untethered operation. The platform enables teleoperation, autonomous SLAM navigation, and vision-based manipulation without external dependencies, providing a low-cost alternative for research and education in robotics and robot learning.

Nikolaus Correll

We compare liquid neural networks with mixture density heads against diffusion policies on Push-T, RoboMimic Can, and PointMaze under a shared-backbone comparison protocol that isolates policy-head effects under matched inputs, training budgets, and evaluation settings. Across tasks, liquid policies use roughly half the parameters (4.3M vs. 8.6M), achieve 2.4x lower offline prediction error, and run 1.8 faster at inference. In sample-efficiency experiments spanning 1% to 46.42% of training data, liquid models remain consistently more robust, with especially large gains in low-data and medium-data regimes. Closed-loop results on Push-T and PointMaze are directionally consistent with offline rankings but noisier, indicating that strong offline density modeling helps deployment while not fully determining closed-loop success. Overall, liquid recurrent multimodal policies provide a compact and practical alternative to iterative denoising for imitation learning.

James Watson, Nikolaus Correll

Uncertainty in perception, actuation, and the environment often require multiple attempts for a robotic task to be successful. We study a class of problems providing (1) low-entropy indicators of terminal success / failure, and (2) unreliable (high-entropy) data to predict the final outcome of an ongoing task. Examples include a robot trying to connect with a charging station, parallel parking, or assembling a tightly-fitting part. The ability to restart after predicting failure early, versus simply running to failure, can significantly decrease the makespan, that is, the total time to completion, with the drawback of potentially short-cutting an otherwise successful operation. Assuming task running times to be Poisson distributed, and using a Markov Jump process to capture the dynamics of the underlying Markov Decision Process, we derive a closed form solution that predicts makespan based on the confusion matrix of the failure predictor. This allows the robot to learn failure prediction in a production environment, and only adopt a preemptive policy when it actually saves time. We demonstrate this approach using a robotic peg-in-hole assembly problem using a real robotic system. Failures are predicted by a dilated convolutional network based on force-torque data, showing an average makespan reduction from 101s to 81s (N=120, p<0.05). We posit that the proposed algorithm generalizes to any robotic behavior with an unambiguous terminal reward, with wide ranging applications on how robots can learn and improve their behaviors in the wild.

Sarah Aguasvivas Manzano, Dana Hughes, Cooper Simpson et al.

We present a library to automatically embed signal processing and neural network predictions into the material robots are made of. Deep and shallow neural network models are first trained offline using state-of-the-art machine learning tools and then transferred onto general purpose microcontrollers that are co-located with a robot's sensors and actuators. We validate this approach using multiple examples: a smart robotic tire for terrain classification, a robotic finger sensor for load classification and a smart composite capable of regressing impact source localization. In each example, sensing and computation are embedded inside the material, creating artifacts that serve as stand-in replacement for otherwise inert conventional parts. The open source software library takes as inputs trained model files from higher level learning software, such as Tensorflow/Keras, and outputs code that is readable in a microcontroller that supports C. We compare the performance of this approach for various embedded platforms. In particular, we show that low-cost off-the-shelf microcontrollers can match the accuracy of a desktop computer, while being fast enough for real-time applications at different neural network configurations. We provide means to estimate the maximum number of parameters that the hardware will support based on the microcontroller's specifications.

Radhen Patel, Rebecca Cox, Branden Romero et al.

We describe the grasping and manipulation strategy that we employed at the autonomous track of the Robotic Grasping and Manipulation Competition at IROS 2016. A salient feature of our architecture is the tight coupling between visual (Asus Xtion) and tactile perception (Robotic Materials), to reduce the uncertainty in sensing and actuation. We demonstrate the importance of tactile sensing and reactive control during the final stages of grasping using a Kinova Robotic arm. The set of tools and algorithms for object grasping presented here have been integrated into the open-source Robot Operating System (ROS).

David Coleman, Nikolaus Correll

We present methods for offline generation of sparse roadmap spanners that result in graphs 79% smaller than existing approaches while returning solutions of equivalent path quality. Our method uses a hybrid approach to sampling that combines traditional graph discretization with random sampling. We present techniques that optimize the graph for the L1-norm metric function commonly used in joint-based robotic planning, purposefully choosing a $t$-stretch factor based on the geometry of the space, and removing redundant edges that do not contribute to the graph quality. A high-quality pre-processed sparse roadmap is then available for re-use across many different planning scenarios using standard repair and re-plan methods. Pre-computing the roadmap offline results in more deterministic solutions, reduces the memory requirements by affording complex rejection criteria, and increases the speed of planning in high-dimensional spaces allowing more complex problems to be solved such as multi-modal task planning. Our method is validated through simulated benchmarks against the SPARS2 algorithm. The source code is freely available online as an open source extension to OMPL.

David Coleman, Ioan Sucan, Sachin Chitta et al.

Developing robot agnostic software frameworks involves synthesizing the disparate fields of robotic theory and software engineering while simultaneously accounting for a large variability in hardware designs and control paradigms. As the capabilities of robotic software frameworks increase, the setup difficulty and learning curve for new users also increase. If the entry barriers for configuring and using the software on robots is too high, even the most powerful of frameworks are useless. A growing need exists in robotic software engineering to aid users in getting started with, and customizing, the software framework as necessary for particular robotic applications. In this paper a case study is presented for the best practices found for lowering the barrier of entry in the MoveIt! framework, an open-source tool for mobile manipulation in ROS, that allows users to 1) quickly get basic motion planning functionality with minimal initial setup, 2) automate its configuration and optimization, and 3) easily customize its components. A graphical interface that assists the user in configuring MoveIt! is the cornerstone of our approach, coupled with the use of an existing standardized robot model for input, automatically generated robot-specific configuration files, and a plugin-based architecture for extensibility. These best practices are summarized into a set of barrier to entry design principles applicable to other robotic software. The approaches for lowering the entry barrier are evaluated by usage statistics, a user survey, and compared against our design objectives for their effectiveness to users.

Carson Kohlbrenner, Niraj Pudasaini, William Xie et al.

Collision-free motion is often aided by tactile and proximity sensors distributed on the body of the robot due to their resistance to occlusion as opposed to external cameras. However, how to shape the sensor's properties, such as sensing coverage; type; and range, to enable avoidant behavior remains unclear. In this work, we present a reinforcement learning framework for whole-body collision avoidance on a humanoid H1-2 robot and use it to characterize how sensor properties shape learned avoidance behavior. Using dodgeball as a benchmark task, we ablate the properties of sensors distributed across the upper body of the robot and find that raw proximity measurements can substitute for explicit object localization provided the sensing range is sufficient and that sparse non-directional proximity signals outpace dense directional alternatives in sample efficiency.

Sarah Aguasvivas Manzano, Patricia Xu, Khoi Ly et al.

We present a high-bandwidth, lightweight, and nonlinear output tracking technique for soft actuators that combines parsimonious recursive layers for forward output predictions and online optimization using Newton-Raphson. This technique allows for reduced model sizes and increased control loop frequencies when compared with conventional RNN models. Experimental results of this controller prototype on a single soft actuator with soft positional sensors indicate effective tracking of referenced spatial trajectories and rejection of mechanical and electromagnetic disturbances. These are evidenced by root mean squared path tracking errors (RMSE) of 1.8mm using a fully connected (FC) substructure, 1.62mm using a gated recurrent unit (GRU) and 2.11mm using a long short term memory (LSTM) unit, all averaged over three tasks. Among these models, the highest flash memory requirement is 2.22kB enabling co-location of controller and actuator.

Khoi Ly, Nicholas Kellaris, Dade McMorris et al.

Soft robotics is a field of robotic system design characterized by materials and structures that exhibit large-scale deformation, high compliance, and rich multifunctionality. The incorporation of soft and deformable structures endows soft robotic systems with the compliance and resiliency that makes them well-adapted for unstructured and dynamic environments. While actuation mechanisms for soft robots vary widely, soft electrostatic transducers such as dielectric elastomer actuators (DEAs) and hydraulically amplified self-healing electrostatic (HASEL) actuators have demonstrated promise due to their muscle-like performance and capacitive self-sensing capabilities. Despite previous efforts to implement self-sensing in electrostatic transducers by overlaying sinusoidal low-voltage signals, these designs still require sensing high-voltage signals, requiring bulky components that prevent integration with miniature, untethered soft robots. We present a circuit design that eliminates the need for any high-voltage sensing components, thereby facilitating the design of simple, low cost circuits using off-the-shelf components. Using this circuit, we perform simultaneous sensing and actuation for a range of electrostatic transducers including circular DEAs and HASEL actuators and demonstrate accurate estimated displacements with errors under 4%. We further develop this circuit into a compact and portable system that couples HV actuation, sensing, and computation as a prototype towards untethered, multifunctional soft robotic systems. Finally, we demonstrate the capabilities of our self-sensing design through feedback-control of a robotic arm powered by Peano-HASEL actuators.

James Watson, Austin Miller, Nikolaus Correll

We present algorithms and results for a robotic manipulation system that was designed to be easily programmable and adaptable to various tasks common to industrial setting, which is inspired by the Industrial Assembly Challenge at the 2018 World Robotics Summit in Tokyo. This challenge included assembly of standard, commercially available industrial parts into 2D and 3D assemblies. We demonstrate three tasks that can be classified into "peg-in-hole" and "hole-on-peg" tasks and identify two canonical algorithms: spiral-based search and tilting insertion. Both algorithms use hand-coded thresholds in the force and torque domains to detect critical points in the assembly. After briefly summarizing the state of the art in research, we describe the strategy and approach utilized by the tested system, how it's design bears on its performance, statistics on 20 experimental trials for each task, lessons learned during the development of the system, and open research challenges that still remain.

Sumeet Batra, John Klingner, Nikolaus Correll

We present a method to register individual members of a robotic swarm in an augmented reality display while showing relevant information about swarm dynamics to the user that would be otherwise hidden. Individual swarm members and clusters of the same group are identified by their color, and by blinking at a specific time interval that is distinct from the time interval at which their neighbors blink. We show that this problem is an instance of the graph coloring problem, which can be solved in a distributed manner in O(log(n)) time. We demonstrate our approach using a swarm chemistry simulation in which robots simulate individual atoms that form molecules following the rules of chemistry. Augmented reality is then used to display information about the internal state of individual swarm members as well as their topological relationship, corresponding to molecular bonds.

Felix von Drigalski, Christian Schlette, Martin Rudorfer et al.

The Industrial Assembly Challenge at the World Robot Summit was held in 2018 to showcase the state-of-the-art of autonomous manufacturing systems. The challenge included various tasks, such as bin picking, kitting, and assembly of standard industrial parts into 2D and 3D assemblies. Some of the tasks were only revealed at the competition itself, representing the challenge of "level 5" automation, i. e., programming and setting up an autonomous assembly system in less than one day. We conducted a survey among the teams that participated in the challenge and investigated aspects such as team composition, development costs, system setups as well as the teams' strategies and approaches. An analysis of the survey results reveals that the competitors have been in two camps: those constructing conventional robotic work cells with off-the-shelf tools, and teams who mostly relied on custom-made end effectors and novel software approaches in combination with collaborative robots. While both camps performed reasonably well, the winning team chose a middle ground in between, combining the efficiency of established play-back programming with the autonomy gained by CAD-based object detection and force control for assembly operations.

Nikolaus Correll, Ray Baughman, Richard Voyles et al.

The Computing Community Consortium (CCC) sponsored a workshop on "Robotic Materials" in Washington, DC, that was held from April 23-24, 2018. This workshop was the second in a series of interdisciplinary workshops aimed at transforming our notion of materials to become "robotic", that is have the ability to sense and impact their environment. Results of the first workshop held from March 10-12, 2017, at the University of Colorado have been summarized in a visioning paper (Correll, 2017) and have identified the key technological challenges of "Robotic Materials", namely the ability to create smart functionality with a minimum of additional wiring by relying on wireless power and communication. The goal of this second workshop was to turn these findings into recommendations for government action. Computation will become an important part of future material systems and will allow materials to analyze, change, store and communicate state in ways that are not possible using mechanical or chemical processes alone. What "computation" is and what is possibilities are, is unclear to most material scientists, while computer scientists are largely unaware of recent advances in so-called active and smart materials. This gap is currently shrinking, with computer scientists embracing neural networks and material scientists actively researching novel substrates such as memristors and other neuromorphic computing devices. Further pursuing these ideas will require an emphasis on interdisciplinary collaboration between chemists, engineers, and computer scientists, possibly elevating humankind to a new material age that is similarly disruptive as the leap from the stone to the plastic age.

Yang Li, Nikolaus Correll

We identify the components of bio-inspired artificial camouflage systems including actuation, sensing, and distributed computation. After summarizing recent results in understanding the physiology and system-level performance of a variety of biological systems, we describe computational algorithms that can generate similar patterns and have the potential for distributed implementation. We find that the existing body of work predominately treats component technology in an isolated manner that precludes a material-like implementation that is scale-free and robust. We conclude with open research challenges towards the realization of integrated camouflage solutions.

Nikolaus Correll, Christoffer Heckman

We posit that embodied artificial intelligence is not only a computational, but also a materials problem. While the importance of material and structural properties in the control loop are well understood, materials can take an active role during control by tight integration of sensors, actuators, computation and communication. We envision such materials to abstract functionality, therefore making the construction of intelligent robots more straightforward and robust. For example, robots could be made of bones that measure load, muscles that move, skin that provides the robot with information about the kind and location of tactile sensations ranging from pressure, to texture and damage, eyes that extract high-level information, and brain material that provides computation in a scalable manner. Such materials will not resemble any existing engineered materials, but rather the heterogeneous components out of which their natural counterparts are made. We describe the state-of-the-art in so-called "robotic materials", their opportunities for revolutionizing applications ranging from manipulation to autonomous driving, and open challenges the robotics community needs to address in collaboration with allies, such as wireless sensor network researchers and polymer scientists.

Yang Li, John Klingner, Nikolaus Correll

We present a distributed algorithm for a swarm of active particles to camouflage in an environment. Each particle is equipped with sensing, computation and communication, allowing the system to take color and gradient information from the environment and self-organize into an appropriate pattern. Current artificial camouflage systems are either limited to static patterns, which are adapted for specific environments, or rely on back-projection, which depend on the viewer's point of view. Inspired by the camouflage abilities of the cuttlefish, we propose a distributed estimation and pattern formation algorithm that allows to quickly adapt to different environments. We present convergence results both in simulation as well as on a swarm of miniature robots "Droplets" for a variety of patterns.

Nikolaus Correll, Prabal Dutta, Richard Han et al.

We describe opportunities and challenges with wireless robotic materials. Robotic materials are multi-functional composites that tightly integrate sensing, actuation, computation and communication to create smart composites that can sense their environment and change their physical properties in an arbitrary programmable manner. Computation and communication in such materials are based on miniature, possibly wireless, devices that are scattered in the material and interface with sensors and actuators inside the material. Whereas routing and processing of information within the material build upon results from the field of sensor networks, robotic materials are pushing the limits of sensor networks in both size (down to the order of microns) and numbers of devices (up to the order of millions). In order to solve the algorithmic and systems challenges of such an approach, which will involve not only computer scientists, but also roboticists, chemists and material scientists, the community requires a common platform - much like the "Mote" that bootstrapped the widespread adoption of the field of sensor networks - that is small, provides ample of computation, is equipped with basic networking functionalities, and preferably can be powered wirelessly.

Dana Hughes, Nikolaus Correll

This paper reviews machine learning applications and approaches to detection, classification and control of intelligent materials and structures with embedded distributed computation elements. The purpose of this survey is to identify desired tasks to be performed in each type of material or structure (e.g., damage detection in composites), identify and compare common approaches to learning such tasks, and investigate models and training paradigms used. Machine learning approaches and common temporal features used in the domains of structural health monitoring, morphable aircraft, wearable computing and robotic skins are explored. As the ultimate goal of this research is to incorporate the approaches described in this survey into a robotic material paradigm, the potential for adapting the computational models used in these applications, and corresponding training algorithms, to an amorphous network of computing nodes is considered. Distributed versions of support vector machines, graphical models and mixture models developed in the field of wireless sensor networks are reviewed. Potential areas of investigation, including possible architectures for incorporating machine learning into robotic nodes, training approaches, and the possibility of using deep learning approaches for automatic feature extraction, are discussed.

Nicholas Farrow, Yang Li, Nikolaus Correll

We present a self-contained, soft robotic hand composed of soft pneumatic actuator modules that are equipped with strain and pressure sensing. We show how this data can be used to discern whether a grasp was successful. Co-locating sensing and embedded computation with the actuators greatly simplifies control and system integration. Equipped with a small pump, the hand is self-contained and needs only power and data supplied by a single USB connection to a PC. We demonstrate its function by grasping a variety of objects ranging from very small to large and heavy objects weighing more than the hand itself. The presented system nicely illustrates the advantages of soft robotics: low cost, low weight, and intrinsic compliance. We exploit morphological computation to simplify control, which allows successful grasping via underactuation. Grasping indeed relies on morphological computation at multiple levels, ranging from the geometry of the actuator which determines the actuator's kinematics, embedded strain sensors to measure curvature, to maximizing contact area and applied force during grasping. Morphological computation reaches its limitations, however, when objects are too bulky to self-align with the gripper or when the state of grasping is of interest. We therefore argue that efficient and reliable grasping also requires not only intrinsic compliance, but also embedded sensing and computation. In particular, we show how embedded sensing can be used to detect successful grasps and vary the force exerted onto an object based on local feedback, which is not possible using morphological computation alone.

Nikolaus Correll, Kostas E. Bekris, Dmitry Berenson et al.

This paper presents a overview of the inaugural Amazon Picking Challenge along with a summary of a survey conducted among the 26 participating teams. The challenge goal was to design an autonomous robot to pick items from a warehouse shelf. This task is currently performed by human workers, and there is hope that robots can someday help increase efficiency and throughput while lowering cost. We report on a 28-question survey posed to the teams to learn about each team's background, mechanism design, perception apparatus, planning and control approach. We identify trends in this data, correlate it with each team's success in the competition, and discuss observations and lessons learned based on survey results and the authors' personal experiences during the challenge.

Lu Ma, Mahsa Ghafarianzadeh, Dave Coleman et al.

We present an unsupervised framework for simultaneous appearance-based object discovery, detection, tracking and reconstruction using RGBD cameras and a robot manipulator. The system performs dense 3D simultaneous localization and mapping concurrently with unsupervised object discovery. Putative objects that are spatially and visually coherent are manipulated by the robot to gain additional motion-cues. The robot uses appearance alone, followed by structure and motion cues, to jointly discover, verify, learn and improve models of objects. Induced motion segmentation reinforces learned models which are represented implicitly as 2D and 3D level sets to capture both shape and appearance. We compare three different approaches for appearance-based object discovery and find that a novel form of spatio-temporal super-pixels gives the highest quality candidate object models in terms of precision and recall. Live experiments with a Baxter robot demonstrate a holistic pipeline capable of automatic discovery, verification, detection, tracking and reconstruction of unknown objects.

David Coleman, Ioan A. Sucan, Mark Moll et al.

We present an experienced-based planning framework called Thunder that learns to reduce computation time required to solve high-dimensional planning problems in varying environments. The approach is especially suited for large configuration spaces that include many invariant constraints, such as those found with whole body humanoid motion planning. Experiences are generated using probabilistic sampling and stored in a sparse roadmap spanner (SPARS), which provides asymptotically near-optimal coverage of the configuration space, making storing, retrieving, and repairing past experiences very efficient with respect to memory and time. The Thunder framework improves upon past experience-based planners by storing experiences in a graph rather than in individual paths, eliminating redundant information, providing more opportunities for path reuse, and providing a theoretical limit to the size of the experience graph. These properties also lead to improved handling of dynamically changing environments, reasoning about optimal paths, and reducing query resolution time. The approach is demonstrated on a 30 degrees of freedom humanoid robot and compared with the Lightning framework, an experience-based planner that uses individual paths to store past experiences. In environments with variable obstacles and stability constraints, experiments show that Thunder is on average an order of magnitude faster than Lightning and planning from scratch. Thunder also uses 98.8% less memory to store its experiences after 10,000 trials when compared to Lightning. Our framework is implemented and freely available in the Open Motion Planning Library.

Timothy M. Caldwell, Dave Coleman, Nikolaus Correll

We present a method for system identification of flexible objects by measuring forces and displacement during interaction with a manipulating arm. We model the object's structure and flexibility by a chain of rigid bodies connected by torsional springs. Unlike previous work, the proposed optimal control approach using variational integrators allows identification of closed loops, which include the robot arm itself. This allows using the resulting models for planning in configuration space of the robot. In order to solve the resulting problem efficiently, we develop a novel method for fast discrete-time adjoint-based gradient calculation. The feasibility of the approach is demonstrated using full physics simulation in trep and using data recorded from a 7-DOF series elastic robot arm.