Neurobionics Lab

Tag: Wearable Robotics

Dephy Ankle Exoskeletons

Published on August 3, 2020

Motivation

Robotic ankle exoskeletons can supplement the user’s biological ankle power with mechanical power from the device. These devices have the potential to restore healthy walking mechanics in individuals with mobility challenges or to augment the performance of able-bodied individuals by enabling them to walk farther, run faster or carry heavier loads. We are interested in how to best control these devices and seek to explore the complex physiological interactions between the exoskeleton and the human user.

Originally designed in the Biomechatronics Group at the MIT Media Lab, we use the Dephy ExoBoot exoskeletons, which are commercially available and manufactured by Dephy, Inc. (Maynard, MA). This system was the first autonomous (untethered) robotic ankle exoskeleton to reduce the energy cost of walking below that of walking without the exoskeleton. 


The Dephy ExoBoot ankle exoskeleton used in the physical experiment. A brushless DC motor mounted on a rigid shank assists the user by generating torque through a belt drive transmission (slack in the image) that applies force on a boot-mounted strut. The exoskeleton is securely attached to the user via a shank attachment that transmits the actuator’s torque to the body.

 Approach

With the development of lightweight, autonomous ankle exoskeletons (like the Dephy ExoBoot), we need to design controllers to assist the user. One commonly defined goal of ankle exoskeletons is to reduce the user’s metabolic cost—but this cost function doesn’t take into account more subjective measures such as pain, comfort, stability, or satisfaction. One way to tease out these subjective elements and how they relate to different control strategies is through user feedback about their preference. We let users self-tune their exoskeleton actuation profiles in 2 dimensions by controlling both the timing and magnitude of peak torque delivery using a touch screen tablet. We are interested in how reliably subjects can tune their own exoskeleton actuation settings in 2 dimensions and arrive at their preferred setting.

We are also investigating basic physiological questions, like evaluating how well users can sense differences in exoskeleton controllers and how well they can sense their own metabolic effort. To characterize this ability, we are calculating the Just Noticeable Difference (JND) of metabolic cost, which is the minimum perceivable change in metabolic cost that can be reliably detected. We use the ExoBoot to repeatedly impose different metabolic costs on test subjects over two-minute intervals. We then measure change in metabolic cost during each interval and ask subjects if they think the current cost is higher than previous one. We then aggregate these responses and obtain each subject’s JND.

Contributors: Leo Medrano, Kim Ingraham, Elliott Rouse

Publications

Medrano, R., Thomas, G. C., & Rouse, E. J. (2020). Methods for Measuring the Just Noticeable Difference for Variable Stimuli: Implications for Perception of Metabolic Rate with Exoskeleton Assistance. In International Conference on Biomedical Robotics and Biomechatronics. doi.org/10.1109/BioRob49111.2020.9224374

Mooney, L. M., Rouse, E. J., & Herr, H. M. (2014). Autonomous exoskeleton reduces metabolic cost of human walking. Journal of NeuroEngineering and Rehabilitation11(1), 151. doi.org/10.1186/1743-0003-11-151

Mooney, L. M., Rouse, E. J., & Herr, H. M. (2014). Autonomous exoskeleton reduces metabolic cost of human walking during load carriage. Journal of NeuroEngineering and Rehabilitation11(1), 80. doi.org/10.1186/1743-0003-11-80

Knee Exoskeleton

Published on August 3, 2020

Motivation

Of the estimated 6.6 million stroke survivors in the United States, almost half report having moderate to severe hemiparesis six months post-stroke. For these individuals, common mobility tasks become difficult or impossible. In particular, transitioning from sitting to standing, which requires large knee torques, takes longer to complete and induces postural instability. People with post-stroke hemiparesis are more likely to fall, and approximately one third of falls occur during transitions such as sit-to-stand. During sit-to-stand transitions, stroke patients heavily favor their non-paretic leg, and show marked joint-level kinetic asymmetries.

Because there is considerable asymmetry in the paretic/non-paretic knee torques, and the knees provide 70% of the work, one strategy to improve sit-to-stand characteristics is to provide knee extension assistance using a powered robotic knee exoskeleton. To perform experiments testing the efficacy of using exoskeletal assistance, we need a device which can appropriately apply assistance.  This device must be capable of accurately applying a torque with a low output impedance

Approach

The chosen design utilizes a series-elastic actuator (SEA) configuration to improve force controllability.  An SEA uses a spring between the motor and load (here the human) to decouple motor inertia, reduce passive output impedance, and allow modulation of motor position (rather than motor torque) to change the output torque.  The result is improved torque control—an important goal of our design.

The primary contribution of this design is the type of spring and the way torque is measured.  We use a fiberglass leaf spring—rather than the more common steel coil or torsion springs—because it is able to store significantly (up to 8x) more energy per mass than spring steel, and is easier to mount. Torque is calculated by measuring spring deflection, and using Hooke’s law for springs (F = kx). We measure spring deflection indirectly, by measuring the angular position of the knee and linear position of the linear guide. The spring stiffness (k) also changes as a function of knee angle, and this relationship was empirically determined.

Contributors: Max Shepherd, Elliott Rouse

Publications

Shepherd, M., & Rouse, E.J. (2017). Design and Validation of a Torque-Controllable Knee Exoskeleton for Sit-to-Stand Assistance. IEEE/ASME Transactions on Mechatronics22(4), 1695-1704. doi.org/10.1109/TMECH.2017.2704521

Shepherd, M.K., & Rouse, E. J. (2016, August). Design and Characterization of a Torque-Controllable Actuator for Knee Assistance during Sit-to-Stand. In 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 2228-2231). IEEE.  doi.org/10.1109/EMBC.2016.7591172

Open Source Leg

Published on August 3, 2020

Motivation

Over one million lower-limb amputees live in the United States, most of whom walk at slower speeds and with less efficient gait. Many research groups around the world are striving to eliminates these differences between people with and without amputations. Research groups across the world are investing significant time, money and effort in designing and building specialized, lab-specific robotic prostheses in order to test control strategies. However, the vast differences in mechanical designs between research groups prevents fair comparison and evaluation of control strategies, limiting the impact on amputee quality of life.

We have developed a new robotic leg that provides a robust, inexpensive, modular, and open-source prosthesis to the research community. The Open Source Leg (OSL) offers a common hardware platform for comparison of control strategies, lowers the barrier to entry for prosthesis research, and enables testing within the lab, community, and at home. Utilization of the OSL has the potential to unite the currently fragmented field of prosthetic leg control and facilitate comparison between control strategies. This standardization of hardware may help researchers move closer to achieving the goal of highly functional robotic prosthetic legs.

Approach

The Open Source Leg features high-torque exterior rotor motors initially developed for the drone industry. The transmission consists of belt drives for high efficiency, low cost, and quiet operation. Additionally, the knee prosthesis gives researchers the option to implement series elasticity and select their desired stiffness level. Moreover, the leg does not require precision-machined components and can be easily assembled or disassembled. Information on how to assemble, control, and use the OSL is open source and can be found here.

Contibutors: Alejandro Azocar, Ung Hee Lee, Elliott Rouse

Publications

Azocar, A. F., Mooney, L. M., Duval, J. F., Simon, A. M., Hargrove, L. J., & Rouse, E. J. (2020). Design and clinical implementation of an open-source bionic leg. Nature Biomedical Engineering 4, 1-13. doi.org/10.1038/s41551-020-00619-3

Azocar, A. F., Mooney, L. M., Hargrove, L. J., & Rouse, E. J. (2018, August). Design and Characterization of an Open-Source Robotic Leg Prosthesis. In 2018 7th IEEE International Conference on Biomedical Robotics and Biomechatronics (BioRob) (pp. 111-118). IEEE. doi.org/10.1109/BIOROB.2018.8488057

Wearable Robotics

Published on November 15, 2019

Wearable robotic devices aim to provide assistance to impaired limbs, augment human ability, or help with the rehabilitation of neuromuscular impairments. For individuals with disabilities, these technologies are especially important as they enable increased functionality and mobility.

Open Source Leg
Variable Stiffness Prosthetic Ankle
Inflatable Actuator

Dephy Ankle Exoskeleton
Running Specific Prosthetic Foot
Series Elastic Actuator Testbench
Knee Exoskeleton
Variable Stiffness Orthosis
Decoupled Energy Storage & Return Prosthetic Ankle