|
| Classifications | Author(s) | Methodology | Advantage(s) | Limitation(s) |
|
| Collision-free approach based on human-in-loop in cyber-physical system (CPS) | Nikolaos N. et al. [9] | The proposed CPS architecture enables collision risk assessment on the scale of millisecond regarding the triggering of safety policies | (i) Low cost and efficient time performance
(ii) Method of human detection and enabling reaction signal | (i) Loading status may impact on robot’s action
(ii) Relationship between response time and image processing is unknown |
| Mohammad S. et al. [10] | The proposed framework from data fusion between laser scanner and IMU to calculate the human-robot minimum distance on-the-fly is investigated | (i) Accuracy and reliability are manageable for collision avoidance success | (i) A limited number of industrial tasks was carried out and residual vibration from robot’s motion which acted on results, was not evaluated |
|
| Collision-free approach based on environmental context-awareness in CPS | Hongyi L. et al. [11] | The context-aware collision-free system with sensing module for path planning based on human pose recognition is introduced | (i) Detecting potential collision and planning a path to reach target simultaneously
(ii) Be able to distinguish human pose and assembly context | (i) Burden computation, an exact human pose recognition algorithm and various assembly sequence are challenges |
| Yingzhong T. et al. [12] | A universal control system which allows an operator to communicate his motions to the robot manipulator, is proposed on the self-adaption workspace mapping method | (i) No need to find parameters by trial-and-error
(ii) It adapts to different operators and robots in workspace | (i) The recognition and classification of human movements were not focused
(ii) The effective distance between operator and system was not discussed. |
| Zong C. et al. [13] | A study for the adaptability of the tracked robot in complex working environment is explored. It includes the mechanical structure, static stability in three terrains for human-robot interaction | (i) Appropriate dimension, good mass distribution and limited velocity is advantageous to maintain the system stability on terrains | (i) Load capacity and financial effectiveness are crucial to investigate |
| Wang H. et al. [14] | The investigation on improving the performance of RFID robot system by anticollision scheme is exemplified. The tag collisions in the current slot are detected by proposed method, then further resolve each small tag collision to enhance system | (i) This approach is beneficial to boost the moving speed and identification reliability of the RFID robots in complex environments | (i) The solution did not fully consider in the context of collaboration, stability or dynamical control
(ii) The phenomenon of weak signal, tags on multi-RFID robots and identification efficiency might occur in practice |
| Zong C. et al. [15] | The dynamic process of climbing stairs for the tracked mobile robot is analysed by reason of on the novelty of mechanical structure and working principle | (i) The feasibility of design and flexible motion could be gained in unknown environment | (i) The advanced algorithms should be considered to deal with different sizes of stairs |
|
| Collision-free approach based on mathematical computing control in CPS | Andrea M. Z. et al. [16] | An avoidance strategy based on depth camera that suggests the robot alternative paths to be traversed, is achieved both collision free and minimum traversing time | (i) This scheme could predict unintended contacts with operator and maintain the speed of robot as much as possible close to its max value | (i) The traffic conditions should be studied more if ethernet connection is utilized |
| Mingcong C. et al. [17] | An artificial potential field-based novel model predictive control path replanner is implemented to provide ample space and sufficient time for proper steering/accelerate/brake in such hazardous scenarios | (i) The proposed method is still deemed capable of tracking aggressive collision-free trajectories while maintaining vehicle stability | (i) A model of four-wheel-independently-actuated is not practical |
| Xiangkun H. et al. [18] | A novel emergency steering control strategy consisting of decision-making layer and motion control layer performs a collision avoidance maneuver | (i) The physical limits of driving actuators and nonlinear factors are included | (i) Human’s behaviour was not concerned |
| Joseph F. et al [17] | A new control structure using model predictive and feedback controller with tire nonlinearities provides firstly collision avoidance ability, then temporarily violate stabilization criteria | (i) All of vehicle’s performance capability are utilized to avoid an accident
(ii) Prioritizing targets are defined | (i) Be inappropriate with cargo-transportation vehicle |
|
| Collision-free approach based on human emotion and activity recognition in CPS | Huang X. et al. [19] | Due to the fundamental emotion modulation theory and the neural mechanisms of generating complex motor patterns, a model of emotion generation and modulation to train a recurrent neural network for robot control to perform goal-directed tasks | (i) The emotion-modulating method is able to control manipulator with higher accuracy and faster learning rate | (i) Various faces with many emotions involve the excellent training process |
| Guo S. et al. [20] | A novel idea for nonlinear multiview Laplacian least squares (MvLL) which construct a global Laplacian weighted graph in order to introduce category discriminant information as well as protect the local neighbourhood information, is proposed. | (i) The effectiveness and robustness of MvLL approach is approved although the tasks of multipose and multifeature are taken | (i) The largest disadvantage of this method is the high time complexity |
| Zheng E. et al. [21] | The new musculoskeletal-based-method manipulated by the electrical Impedance Tomography (EIT) signals for continuously estimating wrist flexion/extension angles is mentioned in the field of wearable robot | (i) The model-based method performed better with small training data sizes | (i) It is limited in extracting motion information on large data and multi-degree-of-freedom |
|
| Collision-free approach based on skill-learning/skill-transfer method in CPS | Su H. et al. [22] | A novel methodology by integrating the cognitive learning techniques and the developed control techniques, which allow the robot to be highly intelligent to learn senior surgeons’ skills and to perform the learned surgical operations in semiautonomous surgery, was discussed | The control algorithm does not only provide the learning ability of the human operation skill from multiple demonstrations on a specific task, but also can transfer the learned motion from open surgery by guaranteeing the robot constraints | More complicated motions entail advanced sensors and burden computation |
| Su H. et al. [23] | An improved human-robot collaborative control scheme was proposed in teleoperated minimally Invasive surgery scenario | The contributions are not only the improvement of surgical task accuracy and robot constraint, but also the computation efficiency without trajectory planning | Adding one more force sensor for external forces sensing should be conceived and performed with the physical interaction between the surgical tip and the organs |
| Su H. et al. [24] | The swivel motion reconstruction approach was applied to imitate human-like behaviour using the kinematic mapping in robot redundancy in this research | The algorithm gains the highly online regression prediction for accuracy enhancement and fast computation | It is important to integrate both human upper limb kinematic models and joint limits |
|
| Collision-free approach based on data fusion for navigation method in CPS | Yang Q. et al. [25] | A novel approach of outdoor localization with 3D-laser scanner is proposed to solve the problem of poor localization accuracy in GPS-denied environments. A path planning strategy based on geometric feature analysis and priority evaluation algorithm is also adopted to ensure the safety and reliability of mobile robot’s autonomous navigation and control | The robot has high accuracy of localization without GPS so that the mean error of position is 0.1 m and the mean error of path angle is 6° in the experiment. The function of obstacle avoidance includes static obstacle avoidance, dynamic vehicle avoidance and going through narrow regions | The mechanical structure of mobile platform does not guarantee to overcome any outdoor environment |
| Sanders D. et al. [26] | The purpose of this investigation is to study the effect on time to complete a task depending on how a human operator interacts with a mobile robot. This kind of interaction utilizes two teleoperated mobile-robot systems, three different ways of interacting with robots and several different environments | It may perform better without a sensor system to assist in more complicated environments. Sometimes, it performs better with a camera mounted on the robot compared with premounted cameras observing the environment | It relies heavily on visual feedback and experienced operators |
| Chen D. et al. [27] | A control framework was used and consisted of two levels: one was a decision level to give the possible positions of all nodes in sensor networks, while the other was a control level to move along the edge of obstacles such that the problem of obstacle avoidance can be transformed into a coordination problem of multiple robots | The proposed control approach can guide the mobile robot to avoid obstacles and deal with the corresponding dynamical events so as to locate all sensor nodes for an unknown wireless network | If the number of obstacles increases, the average localization errors between the actual locations and the estimated locations significantly increase |
|
| Collision-free approach based on trajectory online adaptation in tele-robot | Luo J. et al. [28] | An innovative method of human motion prediction according to an autoregressive model for teleoperation was motivated. The robot’s motion trajectory could be updated in real time through updating the parameters of the proposed model | This approach combines both the prediction of human movement as a feedforward component and a virtual force as feedback | The actively interactive performance of robot as well as the existing trouble in time delay should be considered |
| Jing L. et al. [29] | A hybrid shared control approach based on EMG signals sensor and artificial potential field is invented to avoid obstacles owing to the repulsive force and attractive force from human perception | It provides an alternative force feedback solution along with muscle activation and human’s control of intention and prediction | The robustness and long distance control of the proposed work could be noticed |
|
