**1. Introduction**

The upper extremity is an important part of the human body. Research has found that 80% of severe stroke patients have upper extremity motor dysfunction. It is a relatively feasible and efficient treatment method to perform rehabilitation training by using rehabilitation robot equipment to drive patients. However, in the traditional rehabilitation robot training scheme, the robot usually assists the patient to complete the training action after the specific training process is set [1]. The form of this program is very simple, and patients may feel negative and slack during the training process due to boredom.

Many previous studies have shown that the process of autonomous training by patients is very important. Compared with passive exercise training, the active willingness of patients to participate in training can better promote neurocortical reconstruction and motor function recovery [2]. As a new human-computer interaction method, the brain-computer interface (BCI) can bypass the function of nerve transmission channels and muscle parts, and directly establish information communication channels between the brain and the external environment, and control external devices. The application of BCI in the field of rehabilitation has helped a lot of patients with limb dysfunction to carry out rehabilitation training and accelerate their rehabilitation process. Therefore, in the field of rehabilitation medicine, the study of feasible BCI technology has very important social significance [3–5].

**Citation:** Li, C.; Xu, Y.; He, L.; Zhu, Y.; Kuang, S.; Sun, L. Research on fNIRS Recognition Method of Upper Limb Movement Intention. *Electronics* **2021**, *10*, 1239. https://doi.org/ 10.3390/electronics10111239

Academic Editors: João Paulo Morais Ferreira and Tao Liu

Received: 3 April 2021 Accepted: 15 May 2021 Published: 24 May 2021

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

**Copyright:** © 2021 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/).

At present, many BCI researchers at home and abroad have focused on applying BCI technology to the field of upper limb motor function rehabilitation, and have obtained excellent research results. Anirban of the University of Essex and his research partners successfully developed a hybrid BCI device to control the exoskeleton of the hand in order to overcome the problem of low recognition accuracy in BCI system. The system combines EEG and EMG signals. After the grasping intention of the subject is successfully detected, the exoskeleton will perform finger flexion and extension. Finally, the recognition accuracy of the system reached (90.00 ± 4.86)%, significantly improving the performance of BCI system [6]. Zhai Wenwen hoped to improve the life independence of patients with severe motor dysfunction through BCI technology. The upper-limb movement-related instructions can control the robotic arm to complete the rehabilitation training of the shoulder, wrist and elbow. The recognition accuracy of the system is as high as 93% [7]. Yoshiyuki Suzuki studied the effects of human corticospinal excitability on motor tasks in the process of imagining or observing the upper limbs. The experiments have shown that kinesthetic MI, including visualizing and observing the virtual hand, can cause phase-dependent muscle-specific corticospinal stimulation of wrist muscles that match those in the actual hand [8]. Although they have achieved remarkable research results in the field of sports rehabilitation technology, there are still many key technologies that need to be improved. For example, the recognition accuracy of multi-classification tasks is low, real-time performance needs to be improved, and it is difficult for users to autonomously control the pace of rehabilitation training.

This paper proposes a set of upper limb rehabilitation training robot system based on user spontaneous movement fNIRS-BCI. Four upper limb movement paradigms are designed: Lifting up, putting down, pulling back, and pushing forward. The start of each task and the rest time were all controlled by the subjects autonomously without any prompts from the outside world [9]. A variety of classifiers such as RF and SVM were selected for evaluation, and a high accuracy rate was achieved. Furthermore, the most suitable multiclass recognition algorithm was selected.
