Autonomous manipulation for robots in unstructured environments is challenging due to the technical gap between perception and acting in the physical world. This challenge becomes even more difficult when an object’s motion is constrained in space. This article presents a method to rapidly estimate mechanical constraint models of systems in the environment using physics-informed deep neural networks (PINNs). We develop a single network capable of estimating the motion of constrained mechanisms and present the methods for model training using synthetic data. By leveraging six-dimensional constraints to selectively inform the loss function for different motions within the deep neural network, we achieve high accuracy in estimating the motions for linear and rotation constraints with sub-1 deg error. We experimentally evaluate our model on three real examples, validating the applicability of the approach.

Autonomous robotic manipulation in unstructured environments faces many challenges and is hindered by capabilities that bridge the gap between perception and acting on the world. Action plans that are centric to object motion rather than end-of-arm tooling behavior may aid this. This paper presents an autonomous action planner for a feedback linearizeable system comprised of three base motions that can be leveraged on their own or in combination to give custom motion plans. The optimization routine for the three different types of motion are presented, which are integrated into physics informed neural networks. A component of this is the autonomy that decides how to use these plans independently and in combination. This approach is experimentally demonstrated on autonomous drawer opening, door knob turning, and door knob turning with door opening.

Handling objects with unknown or changing masses is a common challenge in robotics, often leading to errors or instability if the control system cannot adapt in real-time. In this paper, we present a novel approach that enables a six-degrees-of-freedom robotic manipulator to reliably follow waypoints while automatically estimating and compensating for unknown payload weight. Our method integrates an admittance control framework with a mass estimator, allowing the robot to dynamically update an excitation force to compensate for the payload mass. This strategy mitigates end-effector sagging and preserves stability when handling objects of unknown weights. We experimentally validated our approach in a challenging pick-and-place task on a shelf with a crossbar, improved accuracy in reaching waypoints and compliant motion compared to a baseline admittance-control scheme. By safely accommodating unknown payloads, our work enhances flexibility in robotic automation and represents a significant step forward in adaptive control for uncertain environments.

This work presents a coordinated admittance-impedance control architecture for underactutated unmanned aerial manipulators (UAM) that maintains inherent stability during physical contact with the environment. Utilizing dynamically reconfigurable in-flight system parameters, the architecture's intuitive control interface allows aerial manipulation (AM) task execution via human-guided teleoperation or automated maneuvers/trajectories. An admittance-controlled unmanned aerial vehicle (UAV) and impedance-controlled manipulator arm are coordinated by a shared force excitation. The effectiveness of the proposed control architecture is validated in hardware experiments demonstrating smooth transitions between free-flight and environmental contact, along with the ability to maintain constant contact while sliding the UAM's end-effector vertically and horizontally across a surface. Copyright (C) 2024 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)

Autonomous and semi-autonomous robot manipulation systems require fast classification and localization of objects in the world to realize online generation of motion plans and manipulation waypoints in real-time. Furthermore, constraints and estimated plausible motions of objects of interest in space is paramount for autonomous manipulation tasks. For non-grasping tasks like pushing a box or opening an unlatched door, physical properties such as the center of mass and location of constraints like hinges or bearings must be considered. This paper presents a methodology for rapidly inferring constraints and motion plans for objects of interest to be manipulated. This approach is based on a combination of object detection, instance segmentation, localization methods, and algebraically relating different semantically labeled objects. These methods for motion estimation are implemented on a color-depth camera (RGB-D) and a 7 degree-of-freedom serial robot arm. The algorithm's performance is evaluated through different arm poses, assessing both centroid accuracy and estimation speed, and motion estimation performance. Algorithms are tested on an exemplar problem consisting of a block constrained on a dual linear rail system, i.e., constrained linear motion. Experimental results showcase the scalability of this approach to multiple classes with sublinear slowdowns and linear motion plan direction errors as low as 1.23E- 4 [rad]. The manuscript also outlines how these methods for rapid constrained object motion estimation can be leveraged for other applications.

Fluid‐driven artificial muscles exhibit a behavior similar to biological muscles which makes them attractive as soft actuators for wearable assistive robots. However, state‐of‐the‐art fluidic systems typically face challenges to meet the multifaceted needs of soft wearable robots. First, soft robots are usually constrained to tethered pressure sources or bulky configurations based on flow control valves for delivery and control of high assistive forces. Second, although some soft robots exhibit untethered operation, they are significantly limited to low force capabilities. Herein, an electrohydraulic actuation system that enables both untethered and high‐force soft wearable robots is presented. This solution is achieved through a twofold design approach. First, a simplified direct‐drive actuation paradigm composed of motor, gear‐pump, and hydraulic artificial muscle (HAM) is proposed, which allows for a compact and lightweight (1.6 kg) valveless design. Second, a fluidic engine composed of a high‐torque motor with a custom‐designed gear pump is created, which is capable of generating high pressure (up to 0.75 MPa) to drive the HAM in delivering high forces (580 N). Experimental results show that the developed fluidic engine significantly outperforms state‐of‐the‐art systems in mechanical efficiency and suggest opportunities for effective deployment in soft wearable robots for human assistance.

Robot manipulation of the environment often uses force feedback control approaches such as impedance control. Impedance controllers can be designed to be passive and work well while coupled to a variety of dynamic environments. However, in the presence of a high gear ratio and compliance in manipulator links, non-passive system properties may result in force feedback instabilities when coupled to certain environments. This necessitates an approach that ensures stability when using impedance control methods to interact with a wide range of environments. We propose a method for improving stability and steady-state convergence of an impedance controller by using a deep neural network to map a damping impedance control parameter. In this paper, a dynamic model and impedance controlled simulated system are presented and used for analyzing the coupled dynamic behavior in worst case environments. This simulation environment is used for Nyquist analysis and closed-loop stability analysis to algorithmically determine updated impedance damping parameters that secures stability and desired performance. The deep neural network inputs utilized present impedance control parameters and environmental dynamic properties to determine an updated value of damping that improves performance. In a data set of 10,000 combinations of control parameters and environmental dynamics, 20.3% of all the cases result in instability or do not meet convergence criterion. Our deep neural network improves this and reduces instabilities and failed control performance to 2.29%. The design of the network architecture to achieve this improvement is presented and compared to other architectures with their respective performances.

The broad dissemination of unmanned aerial vehicles (UAV s), specifically quadrotor aircraft, has accelerated their successful use in a wide range of industrial, military, and agricultural applications. Research in the growing field of aerial manipulation (AM) faces many challenges but may enable the next generation of UAV applications. The physical contact required to perform AM tasks results in dynamic coupling with the environment, which may lead to instability with devastating consequences for a UAV in flight. Considering these concerns, this work seeks to determine whether off-the-shelf flight controllers for quadrotor UAV s are suitable for AM applications by investigating the passivity and coupled-stability of quad rotors using generic cascaded position-attitude (CPA) and PX4 flight controllers. Using a planar 3-degree of freedom (DOF) linearized state-space model and two high fidelity 6-DOF models with the CPA and PX4 closed-loop flight controllers, passivity is analyzed during free flight, and stability is analyzed when the UAV is coupled to environments with varying degrees of stiffness. This analysis indicates that quadrotors using the CPA and PX4 flight controllers are non-passive (except for the PX4 controller in the vertical direction with certain vehicle parameters) and may become unstable when the UAV is coupled with environments of certain stiffnesses. Similarities between the results from the linearized 3-DOF model and nonlinear 6-DOF models in the passivity analysis suggest that using an analytical, linear approach is sufficient and potentially useful for vehicle geometry and controller design to improve stability for AM applications.

Hydraulic artificial muscles offer unrivaled specific power and power density and are instrumental to the improved performance and success of soft robotics and lightweight mobile applications. This paper addresses the lack of model-based impedance control approaches for soft actuators such as hydraulic artificial muscles. Impedance control of actuators and robotic systems has been proven to be an effective approach for interacting with physical objects in the presence of uncertainty. A sliding mode impedance control approach based on Filippov’s principle of equivalent dynamics is introduced and applied to a hydraulic artificial muscle. A nonlinear lumped parameter model of the system is presented and a sliding mode impedance controller is derived. Experimental results show superior performance using model-based sliding mode impedance control versus a linear impedance control law in both tracking of position and stiffness when disturbances are introduced.

Current approaches to physical security suffer from high false alarm rates and frequent human operator involvement, despite the relative rarity of real-world threats. We present a novel architecture for autonomous adaptive physical security called autonomous detection and assessment with moving sensors (ADAMS). ADAMS is a framework for reducing nuisance and false alarms by placing mobile robotic platforms equipped with sensors outside the normal asset perimeter. These robotic agents integrate sensor data from multiple perspectives over time, and autonomously move to obtain the best new data to reduce uncertainty in the threat scene. Inferences drawn from data fused over time provide ultimate decisions regarding whether to alert human operators. This paper describes the framework and algorithms used in a prototype ADAMS implementation. We describe the results of simulations comparing this framework to alternate paradigms. These simulations show ADAMS has a 4x increase in the range at which threats are identified versus traditional static sensors, and a 5x reduction in false alarms triggered versus frameworks where all sensor detections become alarms, leading to reduced operator load. Further, these simulations show this framework for reacting to new potential threats significantly outperforms methods which merely patrol the site. We also present the results of preliminary hardware trials of an exemplar prototype system, providing limited validation of the simulations in a real-time physical demonstration.

Fluidic braided artificial muscles have been studied for close to seventy years. Their high power-to-weight ratio and force-to-weight ratio make them a desirable actuation technology for compact and lightweight mobile manipulation. Use of hydraulics with fluidic artificial muscles has helped realize high actuation forces with new potential applications. To achieve large actuation forces produced from high internal pressure, artificial muscles operate near the limitations of their mechanical strength. Design improvements and future applications in mechanical systems will benefit from detailed theoretical analysis of the fluidic artificial muscle mechanics. This paper presents the theoretical modeling of a hydraulic artificial muscle, analysis of its mechanics, and experimental results that validate the model. A prototype is analyzed that operates at 14 MPa and can generate up to 6.3 kN of force and a displacement of 21.5 mm. This model promises to be useful for mechanical system design and model-based control.

We describe the development and benchtop prototype performance characterization of a mechatronic system for automatically drilling small diameter holes of arbitrary depth, to enable monitoring the integrity of oil and gas wells in situ. The precise drilling of very small diameter, high aspect ratio holes, particularly in dimensionally constrained spaces, presents several challenges including bit buckling, limited torsional stiffness, chip clearing, and limited space for the bit and mechanism. We describe a compact mechanism that overcomes these issues by minimizing the unsupported drill bit length throughout the process, enabling the bit to be progressively fed from a chuck as depth increases. When used with flexible drill bits, holes of arbitrary depth and aspect ratio may be drilled orthogonal to the wellbore. The mechanism and a conventional drilling system are tested in deep hole drilling operation. The experimental results show that the system operates as intended and achieves holes with substantially greater aspect ratios than conventional methods with very long drill bits. The mechanism enabled successful drilling of a 1/16″ diameter hole to a depth of 9″, a ratio of 144:1. Dysfunctions prevented drilling of the same hole using conventional methods.

Due to their high power density, hydraulic systems are increasingly adapted for human scale devices. For example, commercial and utility electricians use electrohydraulic cutting and crimping tools, rather than human powered tools, to cut and crimp wires that exceed 25mm in diameter. These tools greatly reduce worker-related fatigue and strain-type injuries. To improve electrohydraulic tool technology, there is a need to increase the number of applications from a single battery charge. This paper develops a high fidelity nonlinear lumped parameter model of an electrohydraulic crimping hand tool used by professional electricians. The eleventh-order model can predict tool performance with an average error of 6.9% and 4.4% with respect to the maximum energy consumption and crimp time, respectively. Simulation studies were conducted to investigate reducing the energy consumption of the tool. An independent parameter sweep was performed on the pump piston diameter. The gear ratio was a dependent parameter linked through the maximum motor torque. Increasing the pump piston diameter while increasing the gear ratio was shown to decrease the energy consumption of the tool during crimping applications. Simulations suggest that up to 30% energy can be saved per crimp by increasing the pump piston diameter and gear train ratio.

Abstract Pneumatic double acting cylinders are able to provide inherent stiffness and force control for compliant motion control applications. Impedance control methods allow for a broad spectrum of mechanical properties of actuators to be achieved. The range of this spectrum can be increased by simultaneously controlling the actuator’s inherent stiffness and impedance, a concept explored in this paper. Presented here is a sliding mode impedance and stiffness controller for a servo-pneumatic double acting cylinder. Two proportional servo-valves are employed for simultaneous control of the virtual impedance and inherent stiffness of the pneumatic cylinder. Experimental results of tracking trajectories and contact are reported and discussed with respect to different approaches in the literature.

The high force-to-weight ratios of braided fluidic artificial muscles (AMs) are ideal for human scale and mobile robot applications. Prior modeling efforts focus on the theoretical static characteristics or empirical dynamic models of these actuators when pressurized. This paper develops a comprehensive high fidelity theoretical dynamic model based on first principles for braided pneumatic AMs and presents experimental validation. A novel theoretical model for the nonlinear stiffness is derived to describe the pressure–displacement behavior. The stiffness model, together with friction, damping, and inertia models, forms an equation of motion (EOM) for braided pneumatic AMs. The EOM is coupled with first-order servopneumatic pressure dynamics, resulting in a third-order system model. System model simulations are compared to experimental results of prototypes with nine different geometries. On average, the system model is able to predict the quasi-static displacement within 7% and the dynamic response within 11%. The theoretical model is also benchmarked against a high fidelity curve fit method, with the empirical method showing a 2% improvement in only quasi-static scenarios. The model promises to be useful for mechanical system and model-based control designs.

Overhead line workers have anecdotally reported elevated levels of fatigue in forearm muscles when operating the pistol grip control that maneuvers an aerial bucket on a utility truck. Previous research with surface electromyographic (sEMG) recordings of forearm muscles corroborated these reports of muscle fatigue. A new pistol grip was designed that reduces the applied force by 50% in all directions of movement. In laboratory testing, sEMG signals were recorded from the upper extremity muscles of twenty subjects, who operated a conventional-force pistol grip and the 50% reduced-force control to move a 1/15 scale model of an aerial truck boom. The muscle that resulted in the greatest sEMG activity (extensor digitorum communis (EDC)) was the muscle that workers typically pointed to when they reported forearm muscle fatigue from using the control. The reduced-forced pistol grip decreased EDC sEMG by an average of 5.6%, compared to the conventional control, increasing the maximum endurance time by 38% according to muscle fatigue models. This study was the first to quantify muscular activity of a new aerial bucket pistol grip control and the results show promise for improving the occupational health of electric utility overhead line workers, specifically reducing muscle fatigue. Before the new design of the pistol grip can be commercialized, it must be tested in the field on actual equipment.

A nonlinear lumped-parameter state-space model of a pneumatic artificial muscle that accounts for kinetic friction is developed. Model simulations are reported for square-wave command signals at different frequencies. Comparisons to experimental results demonstrate the fidelity of the model. A new sliding mode control tuning parameter is introduced that increases the gradient of the error dynamic poles of the sliding surface with respect to lower order errors. With this method input-output feedback linearization and model observation are not needed. A third-order integral sliding mode control law exhibits steady-state errors of ±15 μm or less with a maximum error of 0.29 mm or less when tracking a 7th-order square-wave position trajectory with an amplitude of 5.40 mm. This simplified sliding mode control law shows advantages compared to a conventional approach.