Potential future bottlenecks for soft robots and their corresponding solutions

This paper briefly introduces the current situation of soft robots. Then, through the analysis of the current situation, it is concluded that there are two bottlenecks for soft robots, which are material bottleneck and performance bottleneck. In terms of materials, a usable soft robot often requires multiple tasks at the same time, but soft robots lack materials that can meet the needs of multiple tasks at present. In terms of performance, soft robots are different from traditional robots. Soft robots have unlimited degrees of freedom, which will lead to the CPU processing a large amount of data. Therefore, this paper proposes to introduce the concept of robot group into soft robots and solve the problems existing in soft robots by using the characteristics of low CPU and low individual strength requirements of robot groups.


Introduction
With the rapid development of science and technology, the rigid robot has made a huge breakthrough in the accuracy of control, diversity of functions and so on, and has gradually been widely used in a variety of industries.For example, the harvesting robot in agriculture, the robot used for assembly in industry, the surgical robot in medical treatment and the diagnostic robot.But the traditional rigid robot still has many defects.Many rigid robots have low security and poor flexibility because of their rigid structure, which makes it difficult for rigid robots to handle some fragile objects and to use them in spaceconstrained environments.However, people are still looking forward to designing robots that can solve these problems.In order to optimize and solve these problems of rigid robots, research into soft robots with higher safety, flexibility and adaptability has gradually attracted the attention of researchers.Due to their soft materials, soft robots have unlimited degrees of freedom and deformation ability, which can realize active deformation to adapt to harsh environments, enter some gaps that traditional robots cannot enter, and help humans explore more fields that traditional rigid robots cannot explore.However, limited by the materials of soft robots, the soft robots in current research can only achieve a single goal.But the problem that the really powerful robot face is not some single goal.Therefore, this research aims to introduce the characteristics of today's soft robots, explore the possible bottlenecks encountered in the future research of functional and utility soft robots, propose solutions to solve these bottlenecks, and point out the research direction of soft robots.In this way, the soft robot develops in the direction of more powerful functions to meet the functional requirements of human beings for robots.

Current functions of soft robot
In nature, organisms can use their own soft structures to adapt to changing environments.But traditional rigid robots have the problem of insufficient freedom, which severely limits the capabilities of traditional robots.In order to solve the problem, scientists put forward the concept of soft robots based on biological inspiration.With the help of bionics, soft robots have preliminarily realized some functions.Octobot is the world's first fully soft robot designed by Harvard University in the United States [1].With its soft materials, Octobot can complete basic activities such as moving and climbing.By observing the movement characteristics of gastropod mollusks in nature, Wang Jiangbei and Fang Yeyang from Shanghai Jiao Tong University designed a crawling multi-air bag biomimetic soft robot [2].By controlling the gas pressure in multiple connected airbags, the airbags have different degrees of deformation, so that the airbags made of silicone materials have different friction coefficients with the ground, and then generate power through the different friction forces of the front and back airbags to drive the robot forward.A modular soft robot is designed by Fei Yanqiong and Lv Haiyang from Shanghai Jiao Tong University, which is composed of multiple variable spherical module units [3].The robot generates driving force by expanding and contracting each connected sphere at a time to drive the robot forward.There is also a robot gripper designed by Soft Robot Tech Co.,Ltd., which controls the expansion and contraction of air bags of different shapes to grasp.With its highly adaptable soft material, the gripper can realize grasping while ensuring to adapt to the shape of the object, can quickly grasp deformable and fragile items made of different materials.

Bottlenecks that multi-functional soft robot will encounter in the future
Through the analysis of the design of these soft robots, it can be found that there are some defects in the current design of these soft robots.First of all, these soft robots are very dependent on the nature and shape of their materials.Spherical Modular Soft Robot such mentioned above, the robot using the characteristics of the rubber material mechanical properties of complex, made the rubber ball module show a nonlinear relationship between pressure and volume when inflation and deflation, and then through multiple inflation and deflation of the spherical module according to the design sequence to achieve the aim of advancing.However, if the motion technology of the soft robot is to be applied to disaster relief, we need the soft robot to ensure fast movement and good ductility, to ensure enough deformation to enter the narrow gap, and to ensure enough strength not to be destroyed by sharp objects in the ruins, and to ensure that the actuator has enough strength to help the trapped people lift heavy objects while protecting them.Nowadays, common soft robots use materials such as rubber, shape memory alloy SMA, Ion-Exchange Polymer Metal Composite (IPMC), and hydrogels.The properties of these materials are detailed in the following They respond to changes in the environment.Such as PH, temperature, electromagnetic signals and so on These materials have their own advantages, such as rubber's good stretching ability, SMA's ability to remember shapes, IPMC's low power consumption, low density, and hydrogels that can react to the environment.However, it is difficult for these materials to have strong strength and strong reactivity at the same time.As a result, in the face of multi-function realization at the same time, it is necessary for material-related researchers to develop a material with very strong performance to meet the needs of such a multi-function.At the same time, whether these materials can be developed is also unknown.Even if they are developed, they could lead to soft robots that are expensive and difficult to popularize because of their materials.
Modern rigid multifunctional robots have a wealth of functions, such as the robot dog developed by Boston Dynamics, which uses a quadruped robotic arm like a quadruped animal to move.At the same time, there is an extremely long and flexible robotic arm in the head of the robot dog.The main functions of this robot are completed by this robotic arm with three degrees of freedom.Using this robot arm, the robot dog can quickly grasp, lift, carry, place and even drag various objects, turn the handle to open the door, open the valve, and perform other functions.The realization of these functions is mainly attributed to the excellent algorithm of the robot, and the powerful CPU.In order to realize grasping, moving, squeezing, climbing, jumping, growing and other functions, soft robot needs to have unlimited degrees of freedom and the ability of continuous deformation, and these movement and deformation data need to be calculated by the CPU.The amount of data that a soft robot must calculate will grow exponentially.
Therefore, if we want to design a multi-functional soft robot that can be really put into use, the requirements for CPU performance will be greatly improved, and the current household CPU is far from the requirements.And the algorithms that control robot motion generally delegate all decisions to a single core.When the robot is too large, as soft robots have unlimited degrees of freedom, a single core will process a huge amount of data, which leads to overloading and easily hitting the limits of the core's processing capabilities, hampering the search for more sophisticated soft robots.

Solutions to these bottlenecks
Although the CPU requirements of full-fledged soft robots are extremely high, research need not be limited by this.On the contrary, in order to make a further breakthrough in performance and solve the problem of the high cost of traditional rigid robots, researchers proposed the concept of swarm robotics.Robot swarms utilize the biological feature of cooperation of nature swarms, and propose that multiple simple robots can be used to accomplish the goal together instead of a single precision robot, which is called swarm intelligence.This concept can also be applied to the field of soft robots, in which multiple simple CPUs are used to design soft robots with relatively single functions and complete tasks together, so as to solve the problems of expensive materials and the need for high-level CPUs in soft robots.This section will briefly introduce the cooperation of biological communities in nature, and then put forward feasible solutions to utilize swarm intelligence in soft robots, and give suggestions for the development of soft robots today.

Nature Swarms
A common feature of most nature swarms is that each individual in the swam is very weak, but when combined together, the group can easily accomplish goals that are impossible for a single individual.For example, in biofilms composed of bacteria, the single bacteria in the biofilms have almost no survival ability, but when the bacteria gather together, they can accomplish the goals of foraging, resisting foreign enemies, internal division of labor and cooperation, so as to ensure the survival of their own group.The resistance to antibacterial agents of the bacteria in the biofilms is 500 times greater than that of individual bacteria of the same kind [10].The same is true for us humans.In a team, everyone exchanges first-hand information with each other, and the reasonable division of labor and cooperation will be very efficient and accurate to complete the goal.This is why we human beings have developed so far.

The Advantages of swarm intelligence
As an emerging research area, the swarm intelligence has attracted many researchers' attention since the concept was introduced in programs in the 1980s [10].Using swarm intelligence to develop robots' group can complete the task with high efficiency while guaranteeing robustness and low cost [11].Swarm intelligence makes use of the advantages of collective cooperation to complete the task, so that the robot group does not need to use the core for centralized control and provides better solutions to the problems to be solved.These advantages can make up for the shortcomings of the current soft robot research.We can use the characteristics of swarm intelligence, which has low requirements for individuals, to reduce the requirements for soft robot materials.We only need to design materials that can meet the basic needs.At the same time, it can also use its characteristics of eliminating the centralized core and leaving the decision to each individual to reduce the requirements for the core.The computing power of the core only needs to meet the activity needs of a single individual.It can also solve these problems while improving the robustness of the soft robot, so that the soft robot has a stronger ability to survive in abnormal and dangerous situations.This will greatly improve the current status of soft robot research.

Conclusion
This paper briefly introduces and analyzes the current status of soft robot research.It is concluded that the currently designed soft robots generally have the defects of high dependence on materials, and the materials currently used in soft robots are very difficult to meet which makes it difficult to continue to develop high-precision multifunctional soft robots.At the same time, high-precision soft robots require more computation than traditional rigid robots because of their infinite degrees of freedom.Most of the current algorithms rely on a single core, which will process a large amount of data, resulting in an overburden for the core and making it easy to reach the ceiling of the current core processing function, hindering the research of more sophisticated soft robots.Therefore, this paper finds the advantages of swarm intelligence that does not rely on a single core and has low requirements for individuals.It is suggested that these advantages can be used to improve the current research status of soft robots, in the hope of being helpful to the development of soft robots.However, this paper focuses primarily on theory, describing the concepts of soft robot and robot swarm and anticipating the results that can be obtained by combining these two concepts at the theoretical level.However, there is no actual design of robot swarm individuals that can be applied to soft robots, and the specific performance improvement of soft robots can not be given through research after the application of this scheme.Future research will focus on the further exploration of the concept of robot swarm and the design of a robot that can implement the theory, so as to promote the development of soft robots.

Figure 3 .
Figure 3. Robot gripper Designed by Soft Robot Tech Co.Ltd.