**1. Introduction**

Walking is one of the most common behaviors in human daily life, and the ability to walk is an important factor for human beings to live independently. However, neurological diseases such as stroke sequelae and Parkinson's disease can lead to impairment of human motor function and decline in walking ability [1], which can seriously affect the quality of life and health of patients. The World Health Organization survey shows that the incidence of stroke in China ranks first in the world, and stroke is characterized by high incidence rate, high disability rate, high mortality rate, and high recurrence rate [2]. According to the report of the National Bureau of Statistics of China, the elderly population in China will reach 267 million, accounting for 18.9% of the national population in 2022. The accelerated process of aging has increased the number of people suffering from neurological diseases, and the conflict with the lack of medical resources has become a serious problem in the health care system [3]. At present, human beings cannot break the laws of nature to prevent the decline in their own motor functions, and many injuries to the body's motor function

**Citation:** Han, Y.; Liu, C.; Zhang, B.; Zhang, N.; Wang, S.; Han, M.; Ferreira, J.P.; Liu, T.; Zhang, X. Measurement, Evaluation, and Control of Active Intelligent Gait Training Systems—Analysis of the Current State of the Art. *Electronics* **2022**, *11*, 1633. https://doi.org/ 10.3390/electronics11101633

Academic Editor: Dong-Joo Kim

Received: 12 April 2022 Accepted: 19 May 2022 Published: 20 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

are irreversible. It has become one of the urgent problems in society to help the elderly or patients overcome movement disorders, restore their walking function, and improve their daily living ability.

Active intelligent gait training systems are robotic devices that actively interact with human lower limbs to provide support and assistance for the body's motor function. Stateof-the-art rehabilitation training walkers or robotic systems mainly have problems such as poor gait adaptability, inability to quantify and feedback rehabilitation effect, single training strategy, and limited sensor usage environment [4–12]. Facing major national needs and the main economic battlefield, it is of great significance to develop intelligent rehabilitation equipment to realize gait perception, evaluation, and feedback in the home environment, and to help rehabilitation physiotherapists to assist patients in restoring motor function. The gait training system is a large area of research which integrates mechanical design, sensing technology, intelligent control, and robotics technology. At the forefront of the research of intelligent gait training systems and evaluation methods, there are three important basic scientific problems to be solved, which include: (1) the measurement of lower limb movement and the prediction of movement intention, (2) the construction of a walking ability evaluation model based on clinical knowledge and lower limb movement data, and (3) the formulation of personalized gait training strategy of collaborative control of human– machine system. Therefore, the key words "gait measurement and intention recognition", "gait evaluation", and "gait training control strategy" were used in the literature review. Additionally, this review not only selected publications that directly describe or introduce any gait training system, but also retains those publications that focus on any of the three basic scientific problems mentioned above.

In this review, the current active intelligent gait training systems are investigated and discussed from three perspectives, in accordance with three critical scientific problems put forward above, which are measurement and prediction of lower limb movement, evaluation of the effect of gait rehabilitation, and the control strategy of gait training. The main limitations and challenges are then discussed, and potential future directions of intelligent gait training systems are put forward.

#### **2. Human Gait Measurement and Intention Recognition**

#### *2.1. Gait Movement Measurement*

The active intelligent gait training systems have the ability to monitor patient's movement in real time [13]. At present, human's body movements are mainly measured through the fixed force platform and optical motion capture system [14–16] in the gait laboratory, or multiple movement and force sensors worn on the limb [17–19]. The former is highly accurate but limited by the measurement environment, and the latter may interfere with the normal human movement.

The current main human movement measurement methods used in gait training systems are shown in Figure 1. Gait motion measurement techniques used in each of the included studies [14–36] and their characteristics are shown in Table 1. Vision-based methods are one of the important methods for monitoring the posture and movement of the human body and have a wide range of applications [20–23]. Based on the image global joint summation problem or the hierarchical detection fusion problem, deep learning methods have been widely studied for the estimation of human pose [24,25]. However, visual methods have problems such as clothing occlusion, dark environment, high system complexity, difficult installation, and privacy issues, and there are limitations in actual human–machine coordinated movement. The wearable sensing system of human body dynamics analysis consists of multiple sensors, including gyroscopes, pressure sensors, angle sensors, inertial sensors, etc., but it has difficulties in obtaining displacement and relative poses from human to machine. The radio frequency (RF) signal-based method can use the data characteristics of the human body and its motion in the radar image to measure the three-dimensional relative pose and radial velocity [35]. The latest research [36] shows that it has obvious advantages in solving problems such as occlusion and threedimensional reconstruction, but at present it still needs in-depth research on issues such as decoupling RF signals of human and machine movement, fusion understanding based on physical models and data, and generalized measurement of abnormal gait. Therefore, it is necessary to study a new type of non-contact sensing technology solution, combining the kinematics information of the lower limbs and plantar pressure detection to form a composite information perception system to accurately predict the movement trend of the patient's lower limbs and use it to evaluate the patient's health and athletic ability. dimensional reconstruction, but at present it still needs in-depth research on issues such as decoupling RF signals of human and machine movement, fusion understanding based on physical models and data, and generalized measurement of abnormal gait. Therefore, it is necessary to study a new type of non-contact sensing technology solution, combining the kinematics information of the lower limbs and plantar pressure detection to form a composite information perception system to accurately predict the movement trend of the patient's lower limbs and use it to evaluate the patient's health and athletic ability.

angle sensors, inertial sensors, etc., but it has difficulties in obtaining displacement and relative poses from human to machine. The radio frequency (RF) signal-based method can use the data characteristics of the human body and its motion in the radar image to measure the three-dimensional relative pose and radial velocity [35]. The latest research [36] shows that it has obvious advantages in solving problems such as occlusion and three-

*Electronics* **2022**, *11*, x FOR PEER REVIEW 3 of 15

**Figure 1.** State-of-the-art motion measurement techniques used in gait training systems. **Figure 1.** State-of-the-art motion measurement techniques used in gait training systems.


