Physical Human-robot Interaction Research Articles

Learning Ergonomic Control in Human–Robot Symbiotic Walking

This article presents an imitation learning strategy for extracting ergonomically safe control policies in physical human–robot interaction scenarios. The presented approach seeks to proactively reduce the risk of injuries and musculoskeletal disorders by anticipating the ergonomic effects of a robot's actions on a human partner, e.g., how the ankle angle of a prosthesis affects future knee torques of the user. To this end, we extend ensemble Bayesian interaction primitives to enable the prediction of latent biomechanical variables. This methodology yields a reactive control strategy, which we evaluate in an assisted walking task with a robotic lower limb prosthesis. Building upon the learned interaction primitives, we also present a model-predictive control (MPC) strategy that actively steers the human–robot interaction toward ergonomic and safe movement regimes. We compare the introduced control strategies and highlight the framework's ability to generate ergonomic, biomechanically safe assistive prosthetic control. A rich analysis of constrained MPC shows a 20× reduction in the effects of large perturbations on prosthetic control system. We empirically demonstrate a 16% reduction in vertical knee reaction forces in real-world jumping experiments utilizing our control methodology and examine other optimal control strategies in simulated walking experiments.

Characterizing Multidimensional Capacitive Servoing for Physical Human–Robot Interaction

Toward the goal of robots performing robust and intelligent physical interactions with people, it is crucial that robots are able to accurately sense the human body, follow trajectories around the body, and track human motion. This study introduces a capacitive servoing control scheme that allows a robot to sense and navigate around human limbs during close physical interactions. Capacitive servoing leverages temporal measurements from a multielectrode capacitive sensor array mounted on a robot's end effector to estimate the relative position and orientation (pose) of a nearby human limb. Capacitive servoing then uses these human pose estimates from a data-driven pose estimator within a feedback control loop in order to maneuver the robot's end effector around the surface of a human limb. We provide a design overview of capacitive sensors for human–robot interaction and then investigate the performance and generalization of capacitive servoing through an experiment with 12 human participants. The results indicate that multidimensional capacitive servoing enables a robot's end effector to move proximally or distally along human limbs while adapting to human pose. Using a cross-validation experiment, results further show that capacitive servoing generalizes well across people with different body size.

Stable Heteroclinic Channel Networks for Physical Human-Humanoid Robot Collaboration.

Human-robot collaboration is one of the most challenging fields in robotics, as robots must understand human intentions and suitably cooperate with them in the given circumstances. But although this is one of the most investigated research areas in robotics, it is still in its infancy. In this paper, human-robot collaboration is addressed by applying a phase state system, guided by stable heteroclinic channel networks, to a humanoid robot. The base mathematical model is first defined and illustrated on a simple three-state system. Further on, an eight-state system is applied to a humanoid robot to guide it and make it perform different movements according to the forces exerted on its grippers. The movements presented in this paper are squatting, standing up, and walking forwards and backward, while the motion velocity depends on the magnitude of the applied forces. The method presented in this paper proves to be a suitable way of controlling robots by means of physical human-robot interaction. As the phase state system and the robot movements can both be further extended to make the robot execute many other tasks, the proposed method seems to provide a promising way for further investigation and realization of physical human-robot interaction.

Haptics based multi-level collaborative steering control for automated driving

Increasing the capability of automated driving vehicles is motivated by environmental, productivity, and traffic safety benefits. But over-reliance on the automation system is known to cause accidents. The role of the driver cannot be underestimated as it will ultimately be the most relevant aspect for trust building and social acceptance of this technology. Here we introduce a driver-oriented automation strategy to achieve collaborative steering. Our approach relies on three major functionalities: interaction, arbitration, and inclusion. The proposed control strategy is grounded in the concept of shared control enabling driver intervention over the automation without deactivation. Well-defined physical human-robot interaction types are made available with the arbitration strategy. The automated driving trajectory is adapted to include the driver intent into the tactical level of trajectory planning. This enables driver initiated rerouting and consistent coordination of all vehicle actuators. In this way, automated vehicles, which rely on sight only, are augmented with the incorporation of the driver intent. The driver is neither replaced by nor excluded from the automation, rather their role remains active to the benefit of trust building and driving safety.

Embodied Communication: How Robots and People Communicate Through Physical Interaction

Early research on physical human–robot interaction (pHRI) has necessarily focused on device design—the creation of compliant and sensorized hardware, such as exoskeletons, prostheses, and robot arms, that enables people to safely come in contact with robotic systems and to communicate about their collaborative intent. As hardware capabilities have become sufficient for many applications, and as computing has become more powerful, algorithms that support fluent and expressive use of pHRI systems have begun to play a prominent role in determining the systems’ usefulness. In this review, we describe a selection of representative algorithmic approaches that regulate and interpret pHRI, describing the progression from algorithms based on physical analogies, such as admittance control, to computational methods based on higher-level reasoning, which take advantage of multimodal communication channels. Existing algorithmic approaches largely enable task-specific pHRI, but they do not generalize to versatile human–robot collaboration. Throughout the review and in our discussion of next steps, we therefore argue that emergent embodied dialogue—bidirectional, multimodal communication that can be learned through continuous interaction—is one of the next frontiers of pHRI.

Involuntary Stabilization in Discrete-Event Physical Human–Robot Interaction

Robots are used by humans not only as tools but also to interactively assist and cooperate with humans, thereby forming physical human-robot interactions. In these interactions, there is a risk that a feedback loop causes unstable force interaction, in which force escalation exposes a human to danger. Previous studies have analyzed the stability of voluntary interaction but have neglected involuntary behavior in the interaction. In contrast to the previous studies, this study considered the involuntary behavior: a human's force reproduction bias for discrete-event human-robot force interaction. We derived an asymptotic stability condition based on a mathematical bias model and found that the bias asymptotically stabilizes a human's implicit equilibrium point far from the implicit equilibrium point and destabilizes the point near the point. The bias model, convergence of the interaction toward the implicit equilibrium point, and divergence around the point were consistently verified via behavioral experiments under three kinds of interactions using three different body parts: a hand finger, wrist, and foot. Our results imply that humans implicitly secure a stable and close relationship between themselves and robots with their involuntary behavior.

Development of Dummy Based on Impedance Properties of Human Soft Tissue Using a Nonlinear Viscoelastic Model

Securing human safety in a physical human–robot interaction (pHRI) has been extensively studied to allow for the expansion of robotic applications. Dummies with the same impedance properties as those of human soft tissues are suitable substitutes for the validation of the safety assurance of the pHRI experiment. This study demonstrated the development process of dummies with high biofidelity. A psychophysiological experiment was designed to replicate the clamping scenario, and the center of the palm, thenar eminence, and forefinger pad were selected as three soft tissues of the human hand to be impacted to obtain the dynamic response of the contact force. A nonlinear five-element viscoelastic model was developed to quantify the viscoelasticity of the three soft tissues through curve fitting of the contact force. Subsequently, three dummies were fabricated using a polyurethane resin material. The biofidelity of the dummies was examined via a parametric comparison with the soft tissues and curve comparisons to examine whether the steady-state force and the dynamic response of the contact force were located within the corresponding ranges enclosed by the calibration curves of the soft tissues. All three dummies showed high biofidelity in the case where the steady-state force exceeded approximately 5 N, even though they have different nonlinearity of elasticity from that of soft tissues.

Learning Stable Nonlinear Dynamics and Interactive Force-Aware Variable Impedance Control for Robotic Contact Tasks

This paper presents a novel approach for efficient robot programming in tasks such as polishing and grinding, which addresses the challenges of stability and adaptability in robot motion and control. The approach involves learning movement skills and variable impedance control from human demonstrations. Firstly, a non-parametric Gaussian mixture model and globally stable method are proposed for learning robot rhythmic skills with nonlinear dynamics by taking into account the characteristics of robot motion in human-robot physical interaction. Then, adaptive variable impedance control mechanisms are constructed using Gaussian mixture regression, enabling adaptive perception and online control of unknown environments. Finally, a stability-consideration variable impedance controller is designed to minimize the total control-torque while ensuring motion accuracy, which allows efficient adaptation to uncertain environments and complex tasks. The paper provides a novel framework of robot rhythmic skill learning and variable impedance control for robotic complex surface polishing tasks with nonlinear dynamics based on multiple physical human demonstrations, and verifies the proposed methods with a real Franka-robot polishing task. These implementations would be with applications in intelligent manufacturing, robot-assisted rehabilitation.

Iterative assist-as-needed control with skill learning for physical human-robot interaction

Collaborative robots referring to physical human-robot interaction applications have drawn considerable attention. There have been lots of methods based on skill learning to help robots mimic human collaborative behaviors. However, the stiffness profiles generated by these methods only rely on the quality of collected human demonstrations and cannot be adjusted according to detailed interaction tasks. To this end, this paper combines a skill learning method and an iterative assist-as-needed impedance controller. The skill learning method encodes and generates position and stiffness references. The iterative assist-as-needed controller enables the robot to adjust the generated stiffness profile actively according to the human intention. When the human tries to act as an active role in interaction tasks, the controller takes a lower stiffness parameter, which means it can tolerate more deviation from the desired trajectory, and vice versa. Compared to the previous results, on the one hand, this paper employs a skill learning method to generate an initial stiffness profile for the iterative assist-as-needed controller, which enables the robot to mimic human behaviors; on the other hand, the controller proposed in this paper allows the manual design of the damping and mass parameters while the previous controller accepts only the desired stiffness parameter. Finally, one simulation and one experiment are demonstrated to evaluate the performance of the designed controller.

Robust Admittance Control With Complementary Passivity

This paper studies a robust admittance control problem with a passivity requirement for stable and unstable linear time-invariant systems, motivated by control issues originated from physical human-robot interaction. A complementary admittance control structure is proposed and analyzed, revealing that the nominal performance (admittance tracking and passivity) is decoupled from robustness. Simulations on the admittance control for human arm strength augmentation with a passivity requirement validate the proposed controller design.

Composite Learning Adaptive Interaction Control for High-DoF Collaborative Robots*

Variable impedance control is essential for improving safety and naturalness during physical human-robot interaction. However, the robot target impedance is usually difficult to be realized for existing variable impedance control methods. This paper proposes a composite learning adaptive interaction control strategy for robots to achieve target impedance under parametric uncertainty. A multi-mode adaptive control scheme is defined based on weighting functions to ensure smooth mode transitions. A composite learning technique is applied for exact robot modeling online such that target impedance can be achieved under interval excitation that is much weaker than persistent excitation. The proposed method can achieve variable stiffness and damping with guaranteed stable and safe interaction. Experiments on a collaborative robot with 7 degrees of freedom named Franka Emika Panda have validated the effectiveness and superiority of the proposed method.

Passive Decoupled Multitask Controller for Redundant Robots

Kinematic redundancy in robots makes it possible to execute several control tasks simultaneously. As some tasks are usually more important than others, it is reasonable to dynamically decouple them in order to ensure their execution in a hierarchical way or even without any interference at all. The most widely used technique is to decouple the system by feedback linearization. However, this requires actively shaping the inertia and consequently modifying the natural dynamics of the robot. Here, we propose a passivity-based multitask tracking controller that preserves these inertial properties but fully compensates for task-space cross couplings using external-force feedback. In addition, three formal proofs are provided: uniform exponential stability for trajectory tracking, passivity during physical interaction, and input-to-state stability. The controller is validated in simulations and experiments and directly compared with the hierarchical PD+ approach and the feedback linearization. The proposed approach is well suited for safe physical human–robot interaction and dynamic trajectory tracking if measurements or estimations of the external forces are available.

Adaptive Admittance Control for Safety-Critical Physical Human Robot Collaboration

Physical human-robot collaboration requires strict safety guarantees, due to the fact that robots and humans work in a shared workspace. This paper presents a novel control framework to handle safety-critical position-based constraints for human-robot physical interaction. The proposed methodology is based on admittance control, exponential control barrier functions (ECBFs), and quadratic program (QP) to achieve compliance during the force interaction between human and robot, while simultaneously guaranteeing safety constraints. In particular, the formulation of admittance control is formulated as a second-order nonlinear control system, and the interaction forces between humans and robots are regarded as the control input. A virtual force feedback for admittance control is provided in real-time by using the ECBFs-QP framework as a compensator of the external human forces. A safe trajectory is therefore derived from the proposed adaptive admittance control scheme for a low-level controller to track. The main innovation of the proposed approach is the ability to enable the robot to naturally comply with human forces without violating any safety constraints, even when external human forces incidentally force the robot to do so. The effectiveness of our approach is demonstrated in simulation studies on a two-link planar robot manipulator.

Safe Learning-Based Control of Elastic Joint Robots via Control Barrier Functions

Ensuring safety is of paramount importance in physical human-robot interaction applications. This requires both adherence to safety constraints defined on the system state, as well as guaranteeing compliant behavior of the robot. If the underlying dynamical system is known exactly, the former can be addressed with the help of control barrier functions. The incorporation of elastic actuators in the robot's mechanical design can address the latter requirement. However, this elasticity can increase the complexity of the resulting system, leading to unmodeled dynamics, such that control barrier functions cannot directly ensure safety. In this paper, we mitigate this issue by learning the unknown dynamics using Gaussian process regression. By employing the model in a feedback linearizing control law, the safety conditions resulting from control barrier functions can be robustified to take into account model errors, while remaining feasible. In order to enforce them on-line, we formulate the derived safety conditions in the form of a second-order cone program. We demonstrate our proposed approach with simulations on a two-degree-of-freedom planar robot with elastic joints.

A Commentary on Towards autonomous artificial agents with an active self: Modeling sense of control in situated action

Kahl et al., (2022) present a computational model of an autonomous agent implemented with an active self. With ideas based on the Free Energy Principle (Friston and Kiebel, 2009), their model tackles the challenge to unify higher-level cognitive activities and lower-level sensorimotor control as the autonomous agent maintains situational awareness while interacting with the environment. While Kahl et al., (2022) focus on modeling a single agent, we argue that this challenge similarly appears in modeling human–robot interaction (HRI). In this commentary, we discuss how the conceptual framework from Kahl et al., (2022) could inspire unified models of physical and cognitive HRI and how the modeling approach from Kahl et al., (2022) can potentially be applied to anticipatory thinking in robotics to support the human in daily life.

Learning robotic motion with mirror therapy framework for hemiparesis rehabilitation

The number of hemiplegic patients has rapidly increased in recent years, and intelligent robots have high rehabilitation potential. Robotic mirror therapy (RMT) is a promising therapeutic measure for hemiparesis by voluntarily transferring the motion of the healthy limb (HL) to the impaired limb (IL), in which a robot interacts with and assists the IL to mimic the action of the HL to stimulate the active participation of the injured muscles. Nonetheless, complete replication of the HL movement trajectory to the robot without considering the IL muscle strength cannot ensure safety or facilitate rehabilitation. In this study, a learning-based robotic motion generation scheme was developed for RMT. The robot movement trajectories were modeled with dynamic movement primitives (DMPs), and the physical human-robot interaction was formulated as an impedance model coupled with the DMP model. To adapt the robotic motion to different pilots, reinforcement learning was used to optimize the coupled DMP model parameters. The reinforcement learning approach was implemented based on policy improvement and the path integral (PI2) algorithm, and the cost function was designed to simultaneously ensure training safety and enhance muscle strength. The proposed method was validated using a lower-extremity rehabilitation robot with magnetorheological actuators, and the experimental results demonstrated the feasibility and superiority.

Neural-Gas Network-Based Optimal Design Method for ERT-Based Whole-Body Robotic Skin

Electrical resistance tomography (ERT) is an inferential imaging technique that has been utilized to develop large-scale robotic skin due to its scalable and practical properties. The performance of ERT-based sensors has been improved by optimizing the electrode arrangement, but it has relied on a heuristic design due to an absence of quantitative studies on the effects of the electrode arrangement. This article introduces a novel design method to optimize an electrode arrangement for ERT-based robotic skins. The method is based on a neural-gas network, and it finds an optimal design that maximizes the minimum electric current density. The optimal design was comprehensively evaluated using the ERT forward model, and the result revealed that the optimal design achieved the best intrinsic properties in terms of ill-posedness, sensitivity, and spatial discriminability. For validation, we conducted an indentation experiment on ERT-based robotic skin with the optimal design. Although only 30 electrodes were used to cover the 700 cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> sensing area, physical contacts could be localized with an error of 6.6 ± 3.5 mm, and the two-point resolution was adequate for daily tasks. Finally, the developed robotic skin was integrated with a commercial robot arm, and its use in physical human–robot interaction was demonstrated.

Bounded Energy Collisions in Human–Robot Cooperative Transportation

In this article, we introduce the notion of bounded energy collision for physical human–robot collaboration to allow safe interaction with the environment, by keeping the collision impact within predetermined bounds. The proposed method is based on energy dissipation, implemented by a variable damping strategy along the direction of imminent collisions with the environment, and fully complies with any physical human–robot interaction strategy that combines an indirect force control (e.g., impedance control) with an estimation of the human motion intention. Finally, extensive experimental studies clarify and verify the proposed scheme.

Estimation of the Interaction Forces in a Compliant pHRI Gripper

Physical human–robot interaction (pHRI) is an essential skill for robots expected to work with humans, such as assistive or rescue robots. However, due to hard safety and compliance constraints, pHRI is still underdeveloped in practice. Tactile sensing is vital for pHRI, as constant occlusions while grasping make it hard to rely on vision or range sensors alone. More specifically, measuring interaction forces in the gripper is crucial to avoid injuries, predict user intention and perform successful collaborative movements. This work exploits the inherent compliance of a gripper with four underactuated fingers which was previously designed by the authors and designed to manipulate human limbs. A new analytical model is proposed to calculate the external interaction forces by combining all finger forces, which are estimated by using the gripper proprioceptive sensor readings uniquely. An experimental evaluation of the method and an example application in a control system with active compliance have been included to evaluate performance. The results prove that the proposed finger arrangement offers good performance at measuring the lateral interaction forces and torque around the gripper’s Z-axis, providing a convenient and efficient way of implementing adaptive and compliant grasping for pHRI applications.

Human–robot handover with prior-to-pass soft/rigid object classification via tactile glove

Human–robot handovers constitute a challenging and fundamental aspect of physical human–robot interaction. This paper describes the design and implementation of a human–robot handover pipeline in the case in which both soft and rigid objects are passed by the human to the robot. These objects require different profiles of grasping torques by the robot hand fingers, so as to avoid damaging them. As a viable solution to this problem, a tactile glove worn by the human is used to provide real-time information to a deep neural network, which classifies each object as soft or rigid in the pre-handover phase: this information is passed to the robot, which applies the grasping torque profile suitable for the specific type of object. The proposed method is designed and validated based on experiments with eight human participants and 24 objects. The outcomes of these experiments regarding classification accuracy, force and torque profiles, and evaluation of the subjective experiences via questionnaires, are described and discussed.