Model Learning for Control of a Paralyzed Human Arm with Functional Electrical Stimulation

Wolf, Derek; Hall, Zinnia A.; Schearer, Eric M.

doi:10.1109/icra40945.2020.9196992

Cited by 15 publications

(18 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We test the performance of our controller in tracking different trajectories in these tasks. We compare the performance of our controller to that of a PID controller, a conventional method for controlling FES [26], [29]. In single muscle control scenario of vertical motion, the simulation results show that both controllers can track desired trajectories.…”

Section: Discussionmentioning

confidence: 99%

“…Throughout an episode, the fatigue level changes as a function of applied stimulation intensity and the fatigue rate. As a comparison baseline, we also implement a PID controller on all our tasks, due to it being the conventional method used in many FES control applications such as cycling [24], [25] and arm motion [26]. In tasks requiring control of multiple muscles, such as cycling, PID control is used together with a pre-defined active muscle pattern that determines to which muscles the stimulation is applied.…”

Section: Controller Setup and Learningmentioning

confidence: 99%

See 1 more Smart Citation

I am Robot: Neuromuscular Reinforcement Learning to Actuate Human Limbs through Functional Electrical Stimulation

Wannawas,

Shafti,

Faisal

2021

Preprint

View full text Add to dashboard Cite

Human movement disorders or paralysis lead to the loss of control of muscle activation and thus motor control. Functional Electrical Stimulation (FES) is an established and safe technique for contracting muscles by stimulating the skin above a muscle to induce its contraction. However, an open challenge remains on how to restore motor abilities to human limbs through FES, as the problem of controlling the stimulation is unclear. We are taking a robotics perspective on this problem, by developing robot learning algorithms that control the ultimate humanoid robot, the human body, through electrical muscle stimulation. Human muscles are not trivial to control as actuators due to their force production being non-stationary as a result of fatigue and other internal state changes, in contrast to robot actuators which are wellunderstood and stationary over broad operation ranges. We present our Deep Reinforcement Learning approach to the control of human muscles with FES, using a recurrent neural network for dynamic state representation, to overcome the unobserved elements of the behaviour of human muscles under external stimulation. We demonstrate our technique both in neuromuscular simulations but also experimentally on a human. Our results show that our controller can learn to manipulate human muscles, applying appropriate levels of stimulation to achieve the given tasks while compensating for advancing muscle fatigue which arises throughout the tasks. Additionally, our technique can learn quickly enough to be implemented in real-world human-in-the-loop settings.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Controller Setup and Learningmentioning

confidence: 99%

I am Robot: Neuromuscular Reinforcement Learning to Actuate Human Limbs through Functional Electrical Stimulation

Wannawas,

Shafti,

Faisal

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We developed a trajectory optimization routine which accounts for the muscle capabilities of the individual and the dynamics of the arm to find feasible trajectories to a target arm configuration. We compared the performance of controlling the arm along these optimized planned trajectories compared to naive direct-totarget paths using three control structures that are commonly used in FES-driven reaching: a feedback controller [9,14], a feedforward-feedback controller [19], and a model predictive control (MPC) controller [20]. An illustration of our control framework is seen in Fig.…”

Section: Methodsmentioning

confidence: 99%

“…1) Direct-to-target trajectories: Direct-to-target trajectories were defined as a similar approach to the straight-line trajectories followed in previous research [9,14]. We defined a fifthorder polynomial for each joint which began at the starting arm configuration with zero velocity and ended at the target configuration with zero-velocity producing smooth, minimumjerk trajectories similar to natural human reaching [31].…”

Section: Trajectoriesmentioning

confidence: 99%

See 1 more Smart Citation

Data-driven Dynamic Motion Planning for Practical FES-Controlled Reaching Motions in Spinal Cord Injury

Wolf¹,

Bogert²,

Schearer³

2023

Preprint

View full text Add to dashboard Cite

<p>Functional electrical stimulation (FES) is a promising technology for restoring reaching motions to individuals with upper-limb paralysis caused by a spinal cord injury (SCI). However, the limited muscle capabilities of an individual with SCI have made achieving FES-driven reaching difficult. We developed a novel trajectory optimization method that used experimentally measured muscle capability data to find feasible reaching trajectories. In a simulation based on a real-life individual with SCI, we compared our method to attempting to follow naive direct-totarget paths. We tested our trajctory planner with three control structures that are commonly used in applied FES: feedback, feedforward-feedback, and model predictive control. Overall, trajectory optimization improved the ability to reach targets and improved the accuracy for the feedforward-feedback and model predictive controllers (p < 0.001). The trajectory optimization method should be practically implemented to improve the FESdriven reaching performance. </p>

show abstract

Multiple-model iterative learning control with application to stroke rehabilitation

Zhou,

Freeman,

Holderbaum

2025

Control Engineering Practice

View full text Add to dashboard Cite

Model Learning for Control of a Paralyzed Human Arm with Functional Electrical Stimulation

Cited by 15 publications

References 24 publications

I am Robot: Neuromuscular Reinforcement Learning to Actuate Human Limbs through Functional Electrical Stimulation

I am Robot: Neuromuscular Reinforcement Learning to Actuate Human Limbs through Functional Electrical Stimulation

Data-driven Dynamic Motion Planning for Practical FES-Controlled Reaching Motions in Spinal Cord Injury

Multiple-model iterative learning control with application to stroke rehabilitation

Contact Info

Product

Resources

About