**4. Control Strategy of Gait Training Systems**

Traditional walking devices, such as crutches, walkers, wheelchairs, etc., are mostly passive devices, which cannot solve the problem of coordinated dynamic training of body and lower limb muscle when the elderly and patients walk [81]. On the contrary, active intelligent mobility assistance devices interact physically with the human body, as well as coordinated movement, to provide support and assistance for the body's motor function and help the body maintain and restore its motor function to the greatest extent [82,83]. The typical control diagram of active intelligent gait training systems is shown in Figure 2.

2.

exercise detection with multiple measurements. At the same time, the existing simulation software is generally based on a variety of rule constraints such as muscle force-length relationship constraints, muscle force and joint motion coupling constraints, etc., and optimization theories such as minimizing physiological consumption. However, according to the results of human motion modeling and analysis by related researchers [79,80], in patients with gait disorders, it is often difficult to meet the above constraints due to nervemuscle-skeletal damage, and the dynamic representations such as joint torque are affected

Traditional walking devices, such as crutches, walkers, wheelchairs, etc., are mostly passive devices, which cannot solve the problem of coordinated dynamic training of body and lower limb muscle when the elderly and patients walk [81]. On the contrary, active intelligent mobility assistance devices interact physically with the human body, as well as coordinated movement, to provide support and assistance for the body's motor function and help the body maintain and restore its motor function to the greatest extent [82,83]. The typical control diagram of active intelligent gait training systems is shown in Figure

by motion compensation under the condition of external load changes.

**4. Control Strategy of Gait Training Systems** 

**Figure 2.** Typical control diagram of gait training systems. **Figure 2.** Typical control diagram of gait training systems.

A walker is a walking rehabilitation assistive device used to assist users in standing and walking activities which can effectively help users improve their walking ability and is of great significance to a large number of disabled or elderly people. Based on this, intelligent walking rehabilitation assistive robotics devices effectively use the current rapid development of technology to help users break through the original limitations of walking ability to a certain extent and improve their mobility to meet their daily needs. These technologies include mechanical design technology, embedded system technology, sensing and detection technology, automatic control technology, motor control technology, microelectronics technology, interface technology, and software programming. Table 4 shows several gait training systems and their control strategies from recent studies [84– 102]. Pure force/position control means the gait training systems make corresponding operations by detecting human gait events, but do not care about the information of human– machine interaction, while systems with impedance/admittance control strategy use the human–machine interaction information such as interaction force/torque and relative A walker is a walking rehabilitation assistive device used to assist users in standing and walking activities which can effectively help users improve their walking ability and is of great significance to a large number of disabled or elderly people. Based on this, intelligent walking rehabilitation assistive robotics devices effectively use the current rapid development of technology to help users break through the original limitations of walking ability to a certain extent and improve their mobility to meet their daily needs. These technologies include mechanical design technology, embedded system technology, sensing and detection technology, automatic control technology, motor control technology, microelectronics technology, interface technology, and software programming. Table 4 shows several gait training systems and their control strategies from recent studies [84–102]. Pure force/position control means the gait training systems make corresponding operations by detecting human gait events, but do not care about the information of human–machine interaction, while systems with impedance/admittance control strategy use the human– machine interaction information such as interaction force/torque and relative position. Novel human-in-the-loop control represents a large number of control strategies that both recognize the motion intention of humans and detect the human–machine interaction information, described as 'human-in-the-loop' because the information of the human body takes part in both the input and feedback of the closed-loop control. Yu et al. [84] developed an intelligent three-wheeled mobility aid, which is equipped with infrared sensors and laser rangefinders to ensure human–machine–environmental intelligent interaction in motion. Tao et al. [85] studied the intelligent mobility assistance rehabilitation training device for the needs of standing and gait rehabilitation. A standing support and gait training system that maximizes the patient's own rehabilitation exercise ability was developed by using the pressure sensor on the sole of the foot to detect the user's balance or falling state and feeding back the human lower limb joint and muscle force to a load-reducing suspension system. Zhao et al. [88] developed a gait rehabilitation robot to improve the safety and availability of rehabilitation training for patients. A built-in-robot camera was used to obtain leg movement data, and the knee angle was estimated by a New-type ESMF algorithm to deal with the problem of the brief disappearance of the marker point in the field of view.


**Table 4.** Gait training systems and control strategies of each included.

Functional Electrical Stimulation (FES) is a method of applying low-frequency pulsed current or amplifying it through signal-current conversion and then sending it into the human body to produce immediate effects, artificially causing movement in humans who are paralyzed by damage to the central nervous system. Recently, a large number of research studies proposed robotic systems for gait rehabilitation based on FES method [103–105]. Studies have proven that, combined with FES, the assistive torque required of the gait training systems can be reduced and the muscle strength and joint range of motion of the human body can be improved. However, due to the use of electrode pads, this rehabilitation strategy still has problems such as the inability to stimulate deeper muscles or the trauma of electrode implantation in sEMG and iEMG in Section 3.

Locomat is a robotic gait training system. It is used for gait training for patients with abnormal gait caused by brain injury, spinal injury, neurological injury, muscle injury, and orthopedic diseases, and to improve the motor ability of patients with neurological diseases. In the first few generations of prototypes, Locomat also used the common impedance control based on torque feedback [106], but in the latest generations of Locomat Pro, novel control strategy such as automatic gait-pattern adaptation and path control strategy are applied. Locomat Pro can also perform diagnostic evaluation of patients' gait and there are many cases of clinical application [107–109]. However, it is difficult for such a bulky and expensive product to enter millions of households, and the compliance of the control can still be improved. For patients who have lost their mobility due to nerve damage, how to fully mobilize the patient's own movement intention instead of "passive walking" so as to achieve the treatment of nerve injury diseases is a difficult point in the study of the intelligent gait rehabilitation training systems.
