This Rehab Robot Will Challenge You to Tic-Tac-Toe

Physical rehabilitation is not something that anyone does for fun. You do it grudgingly, after an illness or accident, to try and slowly drag your body back toward what it was able to do before. I've been there, and it sucks. I wasn't there for nearly as long as I should have been, however: rehab was hard and boring, so I didn't properly finish it.

Researchers at Ben-Gurion University of the Negev in Israel, led by Shelly Levy-Tzedek, have been experimenting with ways of making rehab a bit more engaging with the addition of a friendly robot arm. The arm can play a mediocre game of tic-tac-toe with you, using cups placed inside a 3x3 square of shelving. While you're focused on beating the robot, you're also doing repetitive reaching and grasping, gamifying all of that upper-limb rehabilitation and making it suck a whole lot less.

The video below shows how the system works, with a Kinova arm taking turns placing cups on the tic-tac-toe grid with a human subject. Using cups was a deliberate choice, because grasping cups (and cup-like objects) is one of those functional movements related to activities of daily living that rehab focuses on restoring. The second video clip shows the same game being played except with colored lights instead of a robot arm— it's less physically interactive, but it's also significantly faster.

Past research has suggested that using the robot arm rather than the lights would be more motivating, lead to increased performance, and imbue a more positive impression of the rehab tasks overall.

In this study, there were two groups of participants: students with an average age of 25 years, and older adults, with an average age of 73 years. Everyone preferred playing with the robot, at least for the first few games, but when asked to play lots of games in a row (10 or more), the younger group changed their preference to playing with the light-up board instead, because it's quicker. The older folks didn't care as much. Everyone also agreed that if they had to pick one system to take home, it would be the robot. I mean, yeah, of course, it's a robot, who wouldn't want to take it home, right?

Using this system for rehab has a few other advantages beyond just making rehab more fun, too:

An important feature of the proposed system is the ability to track the performance of the patients, in terms of success rates, as well as in terms of exact movement patterns. In future elaborations of the setup, this information can be used to monitor patients’ performance in real time, and adjust the game parameters (e.g., timing, or locations selected by the robot) or the feedback that the users receive on the quality of their movements.

For example, the system could recognize that you're having a bit of trouble reaching for specific spots on the board, and change where it places its pieces to either encourage you to keep working on the range of motion required to reach those spots, or give you a bit of a break by making easier to reach spots more appealing moves. It seems like there'd be many more options for cleverness like this with a slightly more complex game: tic-tac-toe is simple enough that it makes for a good demo, but the optimal strategy quickly becomes obvious. And let's be honest— if the robot was trying even a little bit, the best you'd ever be able to manage would be a draw. 

At this point, it's not necessarily clear that the robotic system actually enhances therapeutic value— people seem to like working with it, but the next step here is for the researchers to test it out with stroke patients to see how much of a difference it makes in practice. 

For more details on this research, we spoke with Shelly Levy-Tzedek via email.

IEEE Spectrum: Can you describe how the design of the game is useful for rehabilitation?

Shelly Levy-Tzedek: Patients after stroke often lose the ability to perform certain everyday activities, such as picking up a cup, to drink from. In rehab, they are often required to repetitively perform that basic activity over and over again, to regain the ability to perform it in a functional way. We developed a game with a robotic partner, where, in order to win the game, the participants have to pick up and place a cup many times over. Thus, they end up performing the repetitive (otherwise boring) task that is required in rehab, but in a fun, engaging setup, and with the ability to track their performance during training.

What other rehabilitation games can a system like this be used for?

We have also used this robotic arm to play the "mirror game," in which one partner (person or robot) is leading, and the other is following its movements in space, and then they reverse roles. This work was published in the same issue of RNN and opens the possibility of extending one's range of movement through this game, where the robot can incrementally increase the extent of the movement that it performs, with the person following it closely. 

You mention that "the speed of the system primed the speed of the participants’ movements." Could you elaborate on how and why you think this happens?

Movement priming is a well known phenomenon between humans. If I cross my arms when we talk, you are more likely to cross your arms, for example. This has not been studied much between humans and robots, and the very few studies that were conducted, did not find that a robot affected the human's movement. The reason that they did not find this effect may be because they either used still images, or robots that moved in ways humans don't. We used a physically embodied robot, which performed human-like movements, and found that it primed the movements of the humans that interacted with it. 

This was the case for the other study I mentioned above, with the mirror game. In the mirror game, people made movements more similar to the robot's movements in terms of size, shape and continuity. In the tic-tac-toe (TTT) game, the people (young and old) moved significantly slower when the played against the slow robot (compared to the LED system). It shows that our movements are affected by the entity that we interact with— be it human or robotic, and this should be taken into consideration when designing any human-robot interaction, in medicine, industry, etc., not just in rehab.

Could there be a potential benefit from using a more anthropomorphic robot instead of the robot arm?

Yes, there could be. During this TTT experiment, people commented on how they would rather the robot had a face, or could interact with them verbally. So while there was no direct comparison in an experiment between an anthropomorphic robot and the robotic arm, it did come up as a desired attribute of the robot from the participants.  

In general, how could robots be a useful addition to traditional kinds of rehabilitation therapy?

Robots offer several benefits when used in addition to traditional therapy. They can help in performing repetitive training for many patients without tiring, and can help increase the motivation of patients to perform their exercise regime, which is currently a major problem in rehab. Patients need to perform a lot of self training, but the compliance rates can be as low as 30%. So people are not getting the amount of exercise they need, and motivation plays a big role in this. Gamifying the exercise routine offers an opportunity to increase engagement, and hopefully improve the clinical outcome of patients. Robots can come into the gap in between appointments with the therapist, and help increase the volume of the exercise.

Robotic gaming prototype for upper limb exercise: Effects of age and embodiment on user preferences and movement, by Danny Eizicovitsa, Yael Edana, Iris Tabakb and Shelly Levy-Tzedek from Ben-Gurion University of the Negev in Be'er Sheva, Israel, appears in Restorative Neurology and Neuroscience.