Table 3 Challenges and opportunity/Weaknesses
Type | Keywords of barriers/limitations | Challenges and opportunities/weaknesses |
|---|---|---|
Dataset | Dataset limitation | Large datasets with different categories of clothes should be created to avoid the dataset limitation. More information is given to networks, the easier is for them to produce good results |
Sensor technology | Deep studies on the integration of sensor technology | A multisensory approach should be used to acquire as much as information the robot needs to accomplish its task in real-time (not only use vision but also using force and tactile inputs) |
Perception | Relying only on vision for the detection of clothes [20] | One possibility not to relay only vision is to take an active perception approach, e.g., turn the cloth over or introduce additional slack for perceiving it better. Moreover, tactile sensing could also help in better perception. This kind of approach is used only in few papers as [69] and should be developed more to achieve better results |
No combination between optical flow with forces and 3D information [55] | Feature description combining optical flow with force data should be studied because significant relationships between force data and optical-flow data could improve the success rate and accuracy of failure detection and recovery | |
Manipulation | Robots could learn to infer forces exerted on humans | Robots could learn to infer the forces that people physically feel during robot- assisted dressing to have a more real dressing scenario |
Occlusion of the cloth [52] | To solve the issue of the occlusion of cloth, we could add additional views of it using a hand-mounted camera or putting more cameras around the grasping scenario | |
In many works the authors concentrate only on a single scenario | Systems should not only perform upper or lower body tasks but should do both and their integration could bring balance into the controller. Moreover, our framework could enable robotic assistance in other dressing tasks such as undressing the person | |
The object grasped is unknow by the robot [30] | In principle, the limitation of not knowing the object grasped could be overcome simply by collecting data from many object manipulation scenarios, so as to learn a single model that generalizes effectively across objects. A more nuanced approach might involve correlating the behaviour of objects in the human demonstrations with other previously experienced manipulations, to put them into correspondence and infer the behaviour of an object for which prior experience is unavailable | |
Experimental and simulation phase | A limitation of some studies is that a soft mannequin is used as a subject or simulating only the dressing task | The robot should work with a real person so that researchers can have feedback from them about the force applied by the robot or other problems that can happen during the dressing task to overcome the limitation of using the robot only in simulation |
Neural Network limitations [65] | Comparing neural networks to see the difference between them and find the better and fast approach to accomplish the task | |
Better planning algorithms [25] | Algorithms that consider the limitations of the arms movements of a robot as well as uncertainty in perception would also improve the performance and the safety of the people that are working with a robot | |
Improves manipulators trajectories | Improving manipulator trajectories should be studied in the future to make the robot more user friendly and to reduce the computation time | |
Autonomy of the robot [57] | The robot should be as autonomous as possible to reduce the computation time. For example, when a robot finds a goal infeasible, it should not request the user to reposition his or herself but should recompute autonomously its trajectory | |
Lack of support for deformable objects in most robotic simulators [28] | Create accurate models of deformable object grasping, incorporate it into widely used simulators and release the environments to create a set of benchmark tasks for future research in the domain | |
Use of markers attached to clothes | Rewards calculated based on markers attached to the clothes which is not a real-world scenario | |
Single scenario | Having different scenarios could improve the quality of the robot because it could adapt it to several situations instead of having only a single scenario | |
Legal and safety aspects | No rules can be found in the legal or safety fields of social robotics [50] | The birth of a regulation for social robots could be an important step to overcome the problem that social robots can’t operate in environments with people without a supervision of an operator Moreover, the safety of user should be the foremost priority |