Motivation.
Imagine this–and if you are British or lived in England as much as I did, it will be an easy exercise. As I was saying, imagine walking home in one of those lovely summer evenings where rain is dropping on you and gusts of wind come from every directions. In your mind, a hot toddy and a warm blanket; in your arm, an umbrella. While you attempt to keep the umbrella above your head, blasts of wind may come at any time, from any directions, acting as perturbations that you need to counteract somehow unless you give up and throw the broken umbrella in the first litter bin you find. Let’s focus on that window of time in which you haven’t given up yet, and you are still hoping to get home relatively dry (sigh). When a sudden blast hits your umbrella, two things happen–probably in this order–(i) you swear and (ii) you co-contract your upper-arm muscles to increase stiffness in the direction of the perceived blast in the attempt to keep the umbrella in its upright position. Same happens when you need to do precise movements as in playing Jenga.
What’s a co-contraction, you ask?
Without getting into the physiological details, let’s think of the wrist as a pulley system. The wrist is a wheel. On top of the wheel, your hand is connected and moves according to the motion of the wheel. For simplicity, we will imagine that the hand can only move in one direction, by flexing and extending your wrist. The wheel is connected to two elastic cables (i.e., tendons) which are pulled by two actuators (i.e., your upper-arm muscles). When your muscles are relaxed, the tendons do not act on the wrist which sits in its natural position. External forces will easily move your hand-wrist system out of its natural position. If you pull one of the tendon, by contracting a muscle, the wrist will move accordingly. If you pull both of the tendons, though, something else entirely happens. If both tendons are pulled with the same force, the wrist does not move, but now it is harder for external forces to perturbate the wrist/hand system. This is called a co-contraction, and we use it everyday–holding a powerdrill so to contrast the vibration or carefully remove/place a Jenga piece.
Why is this important to understand?
Now that we know that the muscles contract for two reasons: to move the joint (i.e., wrist) and increase/decrease stiffness, let’s think about how we can use this information. First, you also need to know that we can read muscle activation by placing non-invasive electrodes on the skin. This technology is called surface Electromyography, or sEMG, and researchers have been trying to figure out a way to read such signals and use it to control stuff, such as prosthetics or wearable robots, or even robot that are remotely teleoperated. So far, we have done a good job to recognise some patterns in the muscle activation and map those into defined commands, such as open/close a gripper or moving up and down a robot finger. Yet, decomposing these signals in what is for motion and what is for stiffness is a very hard problem. Note you can move your hand but at the same time also increasing the stiffness to make a more precise motion, or win at Jenga.
Our latest work.
In a collaboration between Prof. Mohan Sridharan‘s IRLab at University of Birmingham (UK), Prof. Dario Farina‘s group at the Imperial College of London (UK), and myself at Heriot-Watt University in Dubai (UAE), our PhD student Laura Ferrante developed a framework for understanding these signals and enable the operator to use contraction and co-contraction to control motion as well as stiffness and damping of a simulated robotic system such as a prosthetic. In our latest publication “Toward Impedance Control in Human-Machine Interfaces for Upper-Limb Prostheses” on IEEE Transactions on Biomedical Engineering, we present a method that enables modulation of the kinematics (i.e., motion) and dynamics properties (i.e., stiffness and damping) of a robotic system from sEMG. This is done by combining an optimised model of the muscle-tendon units in the upper-limb that identifies the human intention–what motion and what stiffness/damping the user wants to achieve in the robotic system–and then maps it into an impedance controller that implements such intention into the robot system. The results are very promising and open up several possibilities to improve the quality of life of amputees or operators that need to work in difficult situations, i.e., fire-fighters wearing a robotic exoskeleton.
Hence, if I have to answer my own question in the title, the answer would be a solid “no”. It is very unlikely that a person wearing a upper-limb prosthesis will beat you at Jenga with the current technology. Yet, we are very proud of this work, which is a first step towards better and more useful robot technologies. We can’t wait to extend this to a real prototype and see it in action. Stay tuned for more news on this.