A Human-in-the-Loop Comparison of Proportional Control and Adaptive Gravity Compensation in Elbow EMG-Driven Exoskeletons

Francesco Pascucci; Eldison Dimo; Davide Costanzi; Mario Sedilesu; Andrea Calanca

cs.ROOct 19, 2026

A Human-in-the-Loop Comparison of Proportional Control and Adaptive Gravity Compensation in Elbow EMG-Driven Exoskeletons

Authors: Francesco Pascucci, Eldison Dimo, Davide Costanzi, Mario Sedilesu, Andrea Calanca

Organizations: University of Verona, Italy · Department of Engineering for Innovation Medicine, University of Verona, Italy. Strada Le Grazie 15, 37134 Verona, Italy.

Abstract

Electromyography (EMG)-driven control enables wearable robots to assist users based on their neuromuscular activity. Upper-limb exoskeletons typically rely on proportional myoelectric control (PMC), which amplifies user effort but may compromise stability at high assistance levels. EMG-driven adaptive gravity compensation (AGC) offers a promising alternative, counterbalancing gravitational torques while continuously adapting to neuromuscular activity. However, these approaches have not been systematically compared in a human-in-the-loop setting. This study evaluates five control algorithms: two PMC strategies at different assistance levels, a purely integral AGC (AGC-PI), a kinematics- and acceleration-aware model-based (MB) AGC (AGC-AW), and a novel MB AGC approach based on Recursive Least Squares (AGC-RLS) for online payload estimation. Fourteen healthy participants performed target-reaching tasks using an active elbow exoskeleton under 0 kg and 2 kg loading conditions. Performance metrics included muscular activity, movement accuracy, task efficiency, and subjective workload. PMC exhibited a trade-off between physical effort and task performance: low assistance levels required greater physical effort under load, whereas high gains impaired task execution. AGC-PI reduced agonist activity but induced antagonist activation. In contrast, both model-based AGC approaches effectively compensated for external payloads, reducing physical and cognitive demands while improving movement accuracy and completion time.

Figures & tables

Fig. 1 : Architecture of an EMG-driven control system, implemented on a custom-built upper-limb exoskeleton prototype.

Fig. 2 : Block diagram of a Proportional Myoelectric Control (PMC) architecture.

Fig. 3 : Block diagram of the Adaptive Gravity Compensation: Pure Integral (AGC-PI) architecture proposed in [ 10 ] .

Fig. 4 : Block diagram of the Adaptive Gravity Compensation: Acceleration-Aware (AGC-AW) architecture proposed in [ 12 ] .

Fig. 5 : Block diagram of the Adaptive Gravity Compensation: Recursive Least Squares (AGC-RLS) architecture.

Fig. 6 : Experimental setup. Left: Participant performing the $2$ $\mathrm{k}\mathrm{g}$ loaded reaching task using a fixed 1-DoF robotic elbow orthosis and a Myo Armband for EMG acquisition. Right: Custom GUI providing real-time visual feedback of targets and forearm position.

Symbol	Description	Value
$k_{p}$	$F_{c}(s)$ proportional gain	$20.0$
$k_{d}$	$F_{c}(s)$ derivative gain	$0.2$
$\alpha$	PMC force-augmentation level	$0.5$ and $0.75$
$\beta$	AGC-PI integral gain	$1.0$
$\eta$	AGC-AW adaptation rate	$1.0$
$\lambda$	AGC-RLS forgetting factor	$0.997$

TABLE I : Control parameters used in the experimental trials.

Metric	Effect	F-Statistic	df	p-value
Biceps Brachii	Algorithm	$31.16$	$(4,52)$	$<0.001$
	Payload	$32.88$	$(1,13)$	$<0.001$
	Alg. $\times$ Payl.	$24.82$	$(4,52)$	$<0.001$
Triceps Brachii	Algorithm	$11.18$	$(4,52)$	$<0.001$
	Payload	$33.30$	$(1,13)$	$<0.001$
	Alg. $\times$ Payl.	$0.70$	$(4,52)$	$0.522$

TABLE II : Two-Way RM ANOVA Results for Muscular Activity and Task Efficiency.

Fig. 7 : Muscular activity (Normalized Mean) for Biceps and Triceps Brachii. Biceps activation is shown for both 0

\mathrm{k}\mathrm{g}

and 2

\mathrm{k}\mathrm{g}

loads, while Triceps activation is aggregated.

	PMC-50	PMC-75	AGC-PI	AGC-AW	AGC-RLS
PMC-50	–	$<0.001$	$0.010$	$0.021$	$<0.001$
PMC-75	$<0.001$	–	$0.026$	$<0.001$	$<0.001$
AGC-PI	$0.010$	$0.027$	–	$<0.001$	$<0.001$
AGC-AW	$0.021$	$<0.001$	$<0.001$	–	$0.002$
AGC-RLS	$<0.001$	$<0.001$	$<0.001$	$0.002$	–

TABLE III : Cross-tabulation of Bonferroni-corrected

p

-values for pairwise comparisons of Overshoot occurrences between control strategies.

Fig. 8 : Number of overshoots observed for each control algorithm. Values are pooled across the 0

\mathrm{k}\mathrm{g}

and 2

\mathrm{k}\mathrm{g}

load conditions, as system stability was not significantly affected by the external weight.

Fig. 9 : Throughput and Completion Time for each control algorithm. Throughput is shown for both 0

\mathrm{k}\mathrm{g}

and 2

\mathrm{k}\mathrm{g}

loads, whereas Completion Time is pooled across conditions.

NASA-TLX Dimension	$\chi^{2}$	p-value	Kendall’s $W$
Mental Demand	$28.24$	$<0.001$	$0.504$
Physical Demand	$18.93$	$<0.001$	$0.338$
Temporal Demand	$15.82$	$0.003$	$0.282$
Performance	$19.79$	$<0.001$	$0.353$
Effort	$18.82$	$<0.001$	$0.336$
Frustration	$20.43$	$<0.001$	$0.365$

TABLE IV : Friedman Test Results for NASA-TLX Subscales (

N=14

df=4

). Effect size is reported as Kendall’s

W

Fig. 10 : NASA-TLX subjective workload scores across the six subscales. Lower scores indicate lower perceived demand, except for Performance, where lower scores indicate better perceived performance.

Upper-Limb Exoskeletons

See all 14 papers

ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance
Aug 6, 2026Yu Chen, Gong Chen, Xiang LiAnkle Exoskeleton Control Feedback Control
Department of Automation, Tsinghua University, China · Shenzhen MileBot Robotics Co., Ltd, China
SoftPINCH: EMG-Driven Soft Exoskeleton Assistance for Finger Flexion and Grasping
Jun 3, 2026Nicklas Nikolaj Grønvall, Magnus Malthe Sigsgaard Nielsen, Xiaofeng Xiong +1EMG-Based Gesture Recognition Robotic Grasping
SDU Soft Robotics, The Maersk Mc-Kinney Moller Institute, University of Southern Denmark, Odense, Denmark
General Performance Guarantee for Human Torque Estimation-Based Task-Agnostic Assistive Exoskeleton Control
Sep 30, 2026Duy Hoang, Bastien Berret, Olivier Bruneau +1Robotic Control Hip Exoskeleton Control
Universit´e Paris-Saclay, CNRS, ENS Paris-Saclay, LMF, 91190 Gif-sur-Yvette, France · Universit´e Paris-Saclay, Inria, CIAMS, 91190 Gif-sur-Yvette, France · Universit´e Paris-Saclay, ENS Paris-Saclay, LURPA, 91190 Gif-Sur-Yvette, France

Aug 6, 2026J/K move · Enter open · S save

ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance

Yu Chen, Gong Chen, Xiang Li

Department of Automation, Tsinghua University, China · Shenzhen MileBot Robotics Co., Ltd, China

Ankle Exoskeleton Control Feedback Control Assistive Robotics Passivity-Based Control Upper-Limb Exoskeletons

Open paper AskPDF arXiv

Providing assistance across diverse movements is a central objective of exoskeletons, and anatomical knowledge can enable responsive support that generalizes across tasks. However, anatomical assistance has mainly been studied for lower-limb exoskeletons, where periodic, weight-bearing motions impose lower demands on torque precision. Extending such assistance to complex, nonperiodic upper-limb movements remains challenging. This paper proposes Anatomical Torque with Passivity-Based Control (ATP) for safe upper-limb exoskeleton assistance. First, a scalable musculoskeletal simulation framework trains a unified reinforcement-learning muscle controller that generalizes across upper-limb movements and generates anatomical reference torques without complex biomechanical computations. Second, an online torque-refinement scheme adapts the reference to diverse movements, suppresses tendon-induced spikes, and incorporates a learned anomaly score for safe and comfortable assistance. Third, an interaction torque controller delivers assistance through a cable-driven compliant exoskeleton without constraining motion to predefined trajectories, while an energy tank preserves passivity with theoretical guarantees on torque tracking and system passivity. Simulations and real-world experiments show accurate tracking of long-duration motion sequences and generalization to real-time human movements. The controller achieves accurate torque tracking while preserving passivity and resumes tracking after energy-tank replenishment. An EMG study with five participants further shows reduced target-muscle activity during static and dynamic tasks compared with gravity compensation and open-loop assistance, with reductions of up to 48% relative to movement without the exoskeleton in a dynamic multi-joint task.

Abstract

Upper-Limb Exoskeletons

Explore similar work

ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance

SoftPINCH: EMG-Driven Soft Exoskeleton Assistance for Finger Flexion and Grasping

General Performance Guarantee for Human Torque Estimation-Based Task-Agnostic Assistive Exoskeleton Control