Research

Research

• At RLC Lab, we strive to come up with novel designs and control strategies for legged robots in order to enable them to work robustly and efficiently in human environments.

• We regard the design of legged robots as closely connected to that of wearable robots –a field that can impact the lives of millions with limited mobility. By constraining the design within an anthropomorphic workspace, novel ideas can be studied and tested on robots before transferring them to powered exoskeletons and prostheses.

• Likewise, studies in control of legged robots can lead to better understanding of human locomotion, which, in turn, can improve the control paradigms designed for exoskeletons and prostheses.

Robot leg design is critical for the success of a legged robot, both in terms of efficiency and control. We proposed a three-stage framework for design of a leg mechanism based on a template:

• Select a template.
• Optimize the mechanism for best placement of the actuators.
• Optimize the actuators.

A high source of energy loss is the so-called "geometric work", which happens when two or more actuators work against each other. We presented a general theorem for avoiding such losses in a mechanism designed based on a reduced-order model and showed the mechanism of ATRIAS can be changed to improve the energy consumption by about 30% .
Actuator design is another factor that can lead to better and more efficient leg design. An optimization framework with a detailed model of electric actuators (considering all relevant components and constraints) was developed, which can ensure the optimal selection of motors and transmissions. As the figure shows, the combination of optimal mechanism design and actuator selection can lead to a more than 50% decrease of total cost of transport compared to a walking experiment with the base system (ATRIAS). This is of significant importance for the untethered operation of legged systems. We still have yet to achieve the remarkable efficiency of human (COT=0.2), but with these optimal design methods and new motor technologies it is not out of reach.
See the following articles for our optimization scheme for further details:

• Rezazadeh, S., Abate, A., Hatton, R.L., and Hurst, J.W., “Robot Leg Design: A Constructive Framework”, IEEE Access, 6(1), pp. 54369-54387, 2018. (PDF)

• Rezazadeh, S. and Hurst, J.W., “On the Optimal Selection of Motors and Transmissions for Electromechanical and Robotic Systems”, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Chicago, IL, USA, 2014.

ATRIAS is a bipedal robot designed based on the concept of Spring-Loaded Inverted Pendulum (SLIP) model of walking and running. ATRIAS participated in the Maximum Mobility and Manipulation Exhibits of DARPA Robotic Challenge (DRC) in Pomona, CA in June 2015. The walking control algorithm was based on generating a highly stable and robust scheme using forced oscillations. Because the only required feedback is time (which can be measured without any noise or delay) the forced-oscillation paradigm introduced in this work led to highly succesful walking (up to 2.2 m/s, which made ATRIAS the fastest untethered bipedal robot at the time) and maintaining stability against strong pushes and kicks, and managing rough terrain. Also, the highly stable and robust forced-oscillation paradigm enabled us to use "add-ons" to the controller; for example, a reflex based obstacle negotiation module which led to successful walking over unexpected step-ups and step-downs (up to 15 cm).

For technical information and mathematical proofs of the controller, please look at the following articles:

• Rezazadeh, S. and Hurst, J.W., “Control of ATRIAS in Three Dimensions: Walking as a Forced-Oscillation Problem”, the International Journal of Robotics Research (IJRR), 39(7), pp. 774-796, 2020.

• Rezazadeh, S., Hubicki, C., Jones, M., Peekema, A., Van Why, J., Abate, A., and Hurst J.W., “Spring-mass Walking with ATRIAS in 3D: Robust Gait Control Spanning Zero to 4.3 KPH on a Heavily Underactuated Bipedal Robot”, ASME Dynamic Systems and Control Conference (DSCC), Columbus, OH, USA, 2015.

Here are a few videos showing ATRIAS's performance:

Designing a prosthetic leg that can help amputees with different weights and demands manage a wide range of tasks can be highly challenging. The knee-ankle prosthetic leg shown in the following videos has been built with the goal of providing high joint torque and power in order to enable different activities for the amputees. One of the important attributes of this design is its high backdrivability, which enables accurate (open-loop) impedance control. We showed that the measured human joints' "quasi-stiffness" can be closely replicated by the leg and with minimal tuning, the subjects can comfortably walk with the leg.

For further information please refer to:

• Elery, T., Rezazadeh, S., Reznick, E., Gray, L., and Gregg, R., "Reducing Transfemoral Amputee Hip Compensations with a Powered Prosthesis: A Case Series", IEEE Transactions on Neural Systems and Rehabilitation Engineering (under review).

• Elery, T.*, Rezazadeh, S.*, Nesler, C., and Gregg, R., “Design and Validation of a Powered Knee-Ankle Prosthesis with High-Torque, Low-Impedance Actuators”, IEEE Transactions on Robotics, 2020.

Powered prostheses have the advantage of being equipped with the actuators which can provide the energy necessary for locomotion. The challenge is how to synchronize their control with human motions. When the prosthesis is transfemoral (i.e. having both knee and ankle joints) the control problem becomes even more challenging.

We have worked on controllers for both rhythmic and volitional tasks through a set of holonomic virtual constraints and a finite state machine. The controller uses only a single feedback signal (the thigh angle of the prosthetic side from an IMU), without any other sensor on the subject (for example another IMU on the sound side), or sensing the ground contact. In the following video, you can see how an amputee subject uses the leg for a variety of tasks, including slow and fast walking, crossing obstacles, kicking a soccer ball, and backward walking. We showed that using the proposed controller, for all of these activities, compensations such as vaulting and hip circumduction significantly improve compared to the subject's everyday passive leg. For more information see:

• Rezazadeh, S., Quintero, D., Divekar, N., Reznick, E., Gray, L., and Gregg, R., “A Phase Variable Approach for Improved Rhythmic and Non-Rhythmic Control of a Powered Knee-Ankle Prosthesis", IEEE Access, 7(1), pp. 109840-109855, 2019. (PDF)

• Quintero, D., Reznick, E., Lambert, D.J., Rezazadeh, S., Gray, L., and Gregg, R., “Intuitive Clinician Control Interface for a Powered Knee-Ankle Prosthesis: A Case Study”, IEEE Journal of Translational Engineering in Health and Medicine, 6(1), pp. 1-9, 2018.

Series Elastic Actuators (SEAs) can be levearged for smaller and more efficient actuators and at the same time help with force control in the joint level. See our NSF grant page where we present how we explore the optimization of SEAs for various tasks and objectives.

Robotics, Locomotion, and Control Lab

Grand Scheme and Broader Impact

Design of Legged Robots

Control of Legged Robots

Design of High-Performance Powered Prostheses

Control of Multi-Joint Prosthetic Legs

Optimization of Series Elastic Actuators