Towards exoskeleton-human symbiosis: investigating how humans interact with an upper-limb robotic exoskeleton – EXOMAN

The EXOMAN project is organized around 4 work packages. The experimental tests will be conducted on the ABLE exoskeleton, a motorized and highly reversible upper limb exoskeleton. First, a technological platform allowing the measurement of an exhaustive set of human motion parameters (kinematic, dynamic and energetic) during the human-exoskeleton interaction will be developed (WP1). This will allow to conduct in-depth experimental analysis and modeling of human motor behavior in interaction with ABLE (WP2). As the adaptation of the human will likely depend on both the control laws applied by the robot and the physical interfacing between the user and the exoskeleton, two complementary research axes will be pursued. On the one hand, high level anthropomorphic control laws will be developed and tested (WP3). On the other hand, the physical interfacing between the human and the exoskeleton will be optimized (WP4). As a reference movement, a standardized applied task will be to help or assist the operator to move his arm as well as a mass, from one point to another, in an intuitive, comfortable, and effortless way. This interdisciplinary project involving human motor control researchers and roboticists will address both fundamental and technological issues to improve our understanding of HEI and enhance the effectiveness of exoskeletons in real-world applications.

The project established a research platform composed of a last generation ABLE 4D exoskeleton robot, an optoelectronic system for kinematic data recording, wireless EMGs, two force sensors placed on the arm and forearm of the robot to measure interaction effort between the user and the robot. The second objective was to study the human motor behavior and its adaptation in several experiments which differ essentially in the type of control laws and physical interfaces considered. Several studies were conducted during the first part of the project, with the agreement of the local ethics committees. The transparency of the exoskeleton has been improved thanks to the precise identification of the exoskeleton parameters. An antigravity mode compensating for the weight of the human arm has also been developed and studied at the methodological and behavioral levels. It shows that users adapt very quickly to this local gravity change and re-optimize their movement based on energy saving principles. Finally, the behavior of users when faced with a full assistance mode but with a voluntarily too low vigor has been investigated, and the results confirm that vigor, i.e., the speed of the resulting movement, is certainly a key factor of human-exoskeleton interaction. Different physical interface solutions were also tested to evaluate their influence on the quality of the interaction and the human motor behavior. In parallel, a first series of experiments was performed on inter-joint coordination at the shoulder and elbow levels following exposure to a viscous force field. The effect of this mode of control was more pronounced in individuals with higher movement vigor. Changes in motor synergies and fluidity of movement were also variable depending on target position. A second set of experiments is underway to evaluate the effects of arm-exoskeleton coupling on the perception of hand-held tools. Preliminary results from this work indicate that the attachments and mass of the exoskeleton do not degrade the perception of the length of the hand-held object. The next phase of experimentation will verify how assistive and corrective modes influence these capabilities. Finally, a theoretical work to define the concept of inter-joint coordination was undertaken. A work was also conducted on the development of a new metric using a robotic formulation to identify joint preferences from kinematic data. An original formulation of an exoskeleton effort control able to manage multiple fixation points has also been implemented in a preliminary work.

The integration of the user's intention in terms of desired trajectory via EMG and/or predictive models will be the main topic in the rest of the project, as well as the creation of adaptive laws to provide a comfortable and adapted assistance to the user's need according to their behavior in the task. The prediction of the desired trajectory and the way of assisting the user along this trajectory will be investigated, with the goal of making these processes as intuitive and useful as possible for the user. The systematic characterization of the effects of different force fields on motor coordination patterns will also be studied. The extension to questions of improving physical ergonomics at work with the exoskeleton will finally be the subject of dedicated work in the last phase of the project.

Verdel, D., Bastide, S., Vignais, N., Bruneau, O., & Berret, B. (2021). An Identification-Based Method Improving the Transparency of a Robotic Upper Limb Exoskeleton. Robotica, 1-18.
Berret, B., Conessa, A., Schweighofer, N., & Burdet, E. (2021). Stochastic optimal feedforward-feedback control determines timing and variability of arm movements with or without vision. PLOS Computational Biology, 17(6), e1009047.
Verdel, D., Bastide, S., Vignais, N., Bruneau, O., & Berret, B. (2022). Human Weight Compensation With a Backdrivable Upper-Limb Exoskeleton: Identification and Control. Frontiers in Bioengineering and Biotechnology (Vol. 9).
Verdel, D., Bastide, S., Bruneau, O., Berret, B., & Vignais, N. Improving and quantifying the transparency of an upper-limb robotic exoskeleton with a force sensor and electromyographic measures. Computer Methods in Biomechanics and Biomedical Engineering. 24(Sup1):S261-S263, 2021.
Vignais, N., Verdel, D., Bastide, S., Bruneau, O., & Berret B. An identification method to improve the transparency of an exoskeleton: development and validation. Proceedings of the Association des Chercheurs en Activités Physiques et Sportives (ACAPS), Montpellier, France, 2021.
Parry, R., Roby-Brami, A., & Jarrassé, N., Upper-limb joint coordination during and after exposure to a robotic exoskeleton: influence of spontaneous movement vigour. Computer Methods in Biomechanics and Biomedical Engineering. 24(Sup1):S269-S271, 2021.
Bastide, S., Adaptation of human motion to new gravito-inertial dynamics induced by interaction with an actuated upper-limb exoskeleton, PhD Thesis, 2021

The overall objective of the EXOMAN project is to improve the symbiosis between humans and exoskeletons. To do so, we aim to advance fundamental knowledge about Human-Exoskeleton Interaction (HEI) by focusing on the human dimension. The potential use of exoskeletons holds much promise in the fields of ergonomics and healthcare, be it for preventing musculoskeletal disorders or overcoming motor deficits. Regardless of the specific end-goal (e.g. augmentation of an operator’s physical capacity or physical assistance for a patient), active exoskeletons may provide a means to assist the movements of a patient/worker with all the advantages offered by robotics: repeatability, accuracy, adaptability. Following an exponential increase in exoskeleton research over the last decade, several types of robotic exoskeletons have been developed. In particular, upper-limb exoskeletons have generated considerable interest, with many potential applications related to reaching and manipulation of objects in clinical and industrial settings. To date however, the generalization of this technology from research into practical applications (i.e. out of the lab) has been limited. Furthermore, the benefits of these devices over existing techniques (e.g. proactive ergonomics or occupational therapy) have not been scientifically established. Clearly, certain aspects of exoskeleton design are limiting their effectiveness and applicability in real life applications. Beyond the inherent technological challenges (actuators, large weight, energy supply...), a fundamental issue is our limited understanding of human motor control in interaction with an exoskeleton. Our research hypothesis is that breakthroughs in robotic exoskeletons will go hand in hand with a better comprehension of the human contribution to HEI. Quantifying and deciphering how humans adapt to moving while wearing an active upper-limb exoskeleton (and why they do so) will thus be a leading theme of this project.
The EXOMAN project will be organized around 4 scientific work packages. Experimental tests will be conducted with ABLE, a highly-reversible upper-limb robotic exoskeleton. Firstly, a technological platform to measure an exhaustive set of human movement parameters (kinematics, dynamics, and energetics) during HEI will be established (work package 1). This will allow conducting exhaustive analyses of human motor behavior in interaction with ABLE (work package 2). Different movement patterns should be observed as a function of the control applied by the exoskeleton and the physical coupling of the person’s upper limb to the exoskeleton. Two complementary research efforts are thus proposed. On the one hand, anthropomorphic high-level control laws for the upper limb exoskeleton will be developed (work package 3). On the other hand, the design of physical interfaces between the person and the exoskeleton will be optimized (work package 4). A standard applied task will be to help/assist an operator to move their limb and a load, from one location to another, in an ecological, comfortable and effortless way.
In summary, this interdisciplinary project involving motor control scientists and roboticists will tackle both fundamental and technological issues to boost HEI research and bring active exoskeletons closer to real world applications.