Human balance recovery strategiesWe combine computational modeling with novel experimental approaches to study how humans maintain balance during walking and standing, and how this is affected by neuromuscular disorders. Using biomechanical metrics like angular momentum and center-of-mass dynamics, we investigate recovery strategies and apply these insights to develop balance-assisting exoskeletons and prosthetics. This work connects fundamental neuroscience with rehabilitation engineering.

Therapeutic and assistive robotics for gait rehabilitationWe have developed and clinically evaluated robotic exoskeletons, e.g LOPES and Symbitron, to train and assist gait in individuals with stroke or spinal cord injury. These devices are force controlled and use adaptive control strategies to support specific subtasks of walking and improve rehabilitation outcomes. Research includes both hardware innovation and control algorithms that mimic human neuromuscular responses. Clinical trials have demonstrated improved mobility and balance recovery in neurological populations.

Therapeutic and assistive devices for the upper extremitiesWearable devices, whether passive or powered, can play a key role in helping people regain movement and coordination after an injury or illness. They can also assist in restoring lost abilities and even reduce pain by providing support during daily activities.
In our work, we design and build a variety of devices, some stationary, others wearable, ranging from simple passive aids to advanced powered systems. Beyond creating hardware, we also study how these active devices should be controlled to make them feel natural and intuitive, and to support effective motor learning.

Powered leg prosthesisMost prosthetic legs available today are passive, meaning they don’t actively assist movement. Actively powered prosthetic legs, however, have the potential to make a big difference. They can help restore natural leg function and reduce the extra strain placed on the remaining joints.
In our research, we work with existing prosthetic devices, both from companies and open-source projects, to develop and test smart control systems. These systems determine how the prosthesis moves and responds. We explore several approaches, including controllers inspired by natural reflexes, rule-based systems, and advanced solutions that use muscle signals (EMG) to create more intuitive movement.

Occupational exoskeletons and exosuitsWearable exoskeletons and exosuits can help reduce strain on the muscles and joints during everyday activities. This can lower the risk of injuries, which are common in jobs that involve heavy lifting or repetitive movements.
Our research focuses on designing new hardware that provides enough support for workers without causing discomfort or limiting their movement. A key challenge we are working on is creating smart control systems that adjust naturally to different tasks and conditions. To make sure these systems work well, we also develop methods to test and compare exoskeletons and exosuits in realistic workplace scenarios.

Actuator designDuring the past two decades, the lab has developed custom actuators for wearable assistive applications. Actuators typically have high reduction ratios and high friction. By including a series-elastic element (a spring), the spring deflection can be controlled as a direct proxy for interaction torque, also allowing actuators to be moved to another location (on the body). Such actuation requires specialized mechanical design, transmission systems (cables, Bowden cables, hydraulics), electronics design and control methods, but achieves high performance closed-loop torque control.

Control of Physical Human Robot InteractionIn wearable robotics applications, naïve force and motion control methods can lead to interaction instability when attached to the human body. We analyze limits of performance, stability, and passivity to ensure unconditional stability of control approaches. Haptic control methods allow for rendering of (time-varying) masses, dampers and compliant elements that prove unconditionally stable and give assistance as needed during rehabilitation

Learning and Optimization-based ControlOnline and offline optimization is used for the control of wearable robots. Optimal trajectory design via nonlinear programming can generate dynamically consistent robust motions for exoskeletons to follow. Similar strategies are applied in non-linear model predictive control to help with human balance and safety of exoskeletons. Fast online optimization deals with redundancy, task overdetermination, and constraints in exoskeletons. Online learning of human-robot dynamics, human controllers and external disturbances allows for optimal adaptability of controllers.