Robot-assisted rehabilitation systems have shown to be helpful in neuromotor rehabilitation because it is possible to deliver interactive and repeatable sensorimotor exercise and monitor the actual performance continuously. Note that it is also essential to distinguish if subject finds the rehabilitation task difficult or easy, since the difficulty level of a task can yield different emotional state, such as excited, bored, over-stressed, etc., at each subject. It is important to adjust the difficulty level of the task to encourage the non-motivated subjects during the therapy. The physiological measurements, which can be obtained from the biofeedback sensors, can be used to estimate the subject's emotional state during the execution of the rehabilitation task. Machine learning methods can be used to classify the emotional state using the features of the biofeedback sensory data. This is explored in this chapter.
TopIntroduction
This chapter presents primarily the current state of art of the robot-assisted rehabilitation systems, achievements and drawbacks in this area by concluding with challenging research points.
Robot-assisted rehabilitation systems, which have been developed since 1998, have shown to be helpful in neuromotor rehabilitation. Robot-assisted rehabilitation systems deliver interactive and repeatable sensorimotor exercise, and monitor the actual performance continuously. During the rehabilitation process, the task difficulty needs to be decided appropriately challenging to obtain optimal performance from the rehabilitation exercise. A rehabilitation task that is too easy or under challenging can be perceived as boring, and an appropriately challenging rehabilitation task can motivate and provide maximum mental engagement for the patients. Mental engagement of patients have shown to be a key factor to improve the outcome of the rehabilitation (Maclean & Pound, 2000). The difficulty level of the rehabilitation task needs to be modified to better suit the patient’s abilities. Motor learning theory states that learning rate increases when the rehabilitation task challenges and excites the subjects (Guadagnoli & Lee, 2004).
Nature has been inspiring the technology for modeling and developing better systems for humanity through ages (Habib, 2011), (Banik et al, 2008a), (Banik et al, 2008b). In human-computer interaction, physiological (affective) computing, which monitors, analyzes and replies to human activity in real time using physiological/biological signals, has opened many potential applications in simulated flight, learning, biomedical engineering etc. (Noval et al, 2012). Physiological computing has also been getting quite interesting in rehabilitation robotics area (Koenig et al., 2011a), (Swagnetr and Kaber, 2013), (Badesa et al, 2013). Note that it is important that the robot-assisted rehabilitation system first detect the emotional state of the patient such as boredom, excitedness or overstress from physiological signals during the execution of the rehabilitation task, and then accordingly adapt the difficulty level of this task. Therefore, the patients will not frustrate or overstressed while executing the rehabilitation tasks. (Koenig et al., 2011a).