Based on the neuroplasticity principle,1 a lower-limb rehabilitation training robot is a kind of automatic equipment that can recover or rebuild neural pathways2,3 for patients with motor dysfunction. The clinical presentation of a spinal cord injury (SCI) or a stroke comprises motor weakness or complete paresis, complete or partial loss of sensory function. The Swedish therapist Brunnstrom proposed the famous six-recovery-stage theory. Based on this theory, the training has different aims in the early stage of rehabilitation (flaccid paralysis stage), middle stage of rehabilitation (spasm stage), and later stage of rehabilitation (recovery stage). One major principle of neurological rehabilitation is that of motor learning. According to the principle of neural plasticity, repetitive and specific training tasks, which make the cerebral cortex learn and store the correct movement patterns, are important and effective. During rehabilitation, patients have to relearn motor tasks in order to overcome disability and limitations in the completion of daily activities. This is the theoretical basis of rehabilitation treatment. For a robot, the control strategy is provided diversely in different stages of rehabilitation to eliminate abnormal movement patterns. In the early rehabilitation, the passive training mode is usually adopted to help patients according to the predetermined trajectory and improve exercise capacity and reduce muscle atrophy. Then the active assist training mode begins for the patients of the middle recovery stage with moderate strength and relieving muscle spasm. In the later rehabilitation stage, the active resist training mode can be used to encourage patients to participate initiatively. The effect and importance of rehabilitation robots have been internationally recognized.4–8
Giving different state of an illness exhibited by hemiparetic individuals and the different training modes as mentioned above, the gait rehabilitation training robot primarily entails customized designing the parameters including movement trajectory, training speed and strength, and real-time perceiving, adjusting, and controlling. Lower-limb exoskeleton mechanism features of many degrees of freedom, together with the individual and condition differences of patients, so the problems are highlighted about how to accurately plan the correct gait trajectory and how to adjust training modes on time according to the progression. These issues become one of the research foci and technical difficulties of rehabilitation robot.
Most of the typical lower-limb rehabilitation robots in the world are autonomously controlled. The gait training mode planning for them is summarized in two methods, that is, preselected by a physiotherapist and dynamically adjusted by the algorithm. For some representative examples, the horizontal rehabilitation training robot Motion Maker9 can automatically guide patients along a preselected trajectory to perform passive flexion movement training on hip, knee, and ankle joints. The Lokomat10–12 is a kind of body-weight-supported treadmill training (BWSTT) robot that adjusts the assisted power or reference trajectory by the impedance algorithm according to the patient interaction force. Patients can be made available to active and passive training mode. In the case of the lower extremity powered exoskeleton (LOPES) gait rehabilitation robot,13,14 limb reference trajectory is generated by instantaneous mapping with the healthy limb movement. The feasibility and functional improvements achieved in response to the emergence of such self-control rehabilitation robot; however, the existing technological bottleneck is obvious, that is, the limited adaptability of training mode.
The objective of this research is to develop a gait trajectory teaching device, with which the physiotherapist can directly and professionally teach to the robot and therefore present complex actions and adjust training modes as needed. Through master–slave teaching method, such system may provide the adaptability of the robot-mediated training and improve the treatment quality and efficiency, and decrease the difficulties in control algorithm study and the contradiction between the complex algorithms and real-time control.
Because of the more elaborate actions of the upper extremity and hand, teaching and playback technology is first applied to upper-limb rehabilitation training robot, for example, the flexible force feedback master–slave exoskeleton manipulator developed by America General Electric Company,15 the wearable master–slave training equipment of upper limbs driven by pneumatic artificial muscles in Okayama University in Japan,16 and the remotely operated upper-limb training robot of Southeast University in China.17 But there are fewer applications for lower-limb rehabilitation training. A single-joint ankle-foot orthoses designed by Canada, the Centre for Interdisciplinary Research in Rehabilitation and Social Integration is introduced in the literature.18The main cylinder driven by a motor controlled the slave cylinder to drive the orthoses. As described in a literature,19 a wearable master–slave lower-limb training robot driven by pneumatic artificial muscle achieves the teaching and training for the knee and ankle rehabilitation by sensors feeding back the trainer joint torque to the main control mechanism. In most of the studies mentioned above, the limitations existing in master–slave teaching for the lower-limb rehabilitation training robot can be summarized as follows: (1) the teaching device has the characteristics of complex structure, large quality and high inertia, so the physiotherapist is laborious and feels fatigue quickly, (2) the coordinate of the multi joints is demanded highly which may lead to the insufficient operating smoothness of the device, and (3) the feedback joint torque cannot be directly perceived by the physiotherapist but only as the control signal for the device.
In light of the above limitations, a novel multi-joint wearable teaching device is developed with adjustable operating force, which is exerted by light film cylinders. Based on the gravity compensation control method, a physiotherapist operates the teaching device to plan training trajectory smoothly and comfortably while also perceive the scene interaction force came from patients. In this manner, our research solved the existing problems, namely, the weight, the difficult manipulation of the teaching and the less force feedback to the physiotherapist. He operates the teaching device with the master–slave mode may provide various training modes fast and intuitively. The force telepresence from patients makes physiotherapist better controlling the training intensity and realizing the individual rehabilitation training consultation.
In this article, we elaborate five major contents that have been derived from this research as follows: master–slave teaching system solution, structural design of the multi-joint wearable master teaching device, operating force regulation principle and gravity compensation method, operating force regulation performance experiments, and master–slave control experiments.