Soft Robotics: Academic Insights and Perspectives Through Bibliometric Analysis

Abstract Soft robotics is of growing interest in the robot community as well as in public media, and there is an increase in the quality and quantity of publications related to this topic. To formally elaborate this growth, we have used a bibliometric analysis to evaluate the publications in the field from 1990 to 2017 based on the Science Citation Index Expanded database. We present a detailed overview and discussion based on keywords, citation, h-index, year, journal, institution, country, author, and review articles. The results show that the United States takes the leading position in this research field, followed by China and Italy. Harvard University has the most publications, high average number of citations per publication and the highest h-index. IEEE Transactions on Robotics ranks first among the top 20 academic journals publishing articles related to this field, whereas Soft Robotics holds the top position in journals categorized with “ROBOTICS.” Actuator, fabrication, control, material, sensing, simulation, bionics, stiffness, modeling, power, motion, and application are the hot topics of soft robotics. Smart materials, bionics, morphological computation, and embodiment control are expected to contribute to this field in the future. Application and commercialization appear to be the initial driving force and final goal for soft robots.


Introduction
I n recent years, soft robotics has become one of the fastest growing topics in the robotic community, and its rise in academia 1 suggests the potential to revolutionize the role of robotics in society and industry. 2 Despite this prodigious future, the research field is quite young. According to a survey of the literature, the term ''soft robot'' was first used for a rigid pneumatic hand, 3 which had a certain degree of object compliance owing to the compressibility of gas. Afterward, soft robot was gradually used in a variety of articles, patents, reports, and other scientific documents, yet still represented a robot or similar machine composed of rigid materials. In 2008, the term ''soft robotics'' was adopted to describe investigation of rigid robots with compliant joints, 4 as well as soft material-based robots with large scale flexibility, deformability, and adaptability. 5 But the efforts to invent new robots that are totally different from their conventional rigid counterparts really started far before the appearance of the professional terminology. In the 1950s, McKibben developed braided pneumatic actuators for an orthotic appliance for polio patients. 6 The McKibben artificial muscle was widely investigated and employed in different types of robot design. 7 In 1990, Shimachi and Matumoto 8 reported their work on soft fingers. One year later, Suzumori et al. 9 published their flexible microactuator made of silicone rubber and tried several applications. [10][11][12] In the following dozen years, similar structures were developed, named as pneumatic bellows actuators, 13 electrostrictive polymer artificial muscle actuators, 14 rubbertuator, 15 fluidic muscle, 16 pneumatic rotary soft actuator, 17 flexible fluidic actuator, 18 flexible pneumatic actuator, 19 tentacle manipulator, 20 elephant trunk manipulator, 21 Air-Octor, 22 OctArm, 23 caterpillar robot, 24 Clobot, 25 continuum manipulators, 26 and so on. Despite the different mechanisms, structures, and motion performance, these actuators and devices are clearly key developments in the discipline of soft robotics. 27,28 Although soft robotics has a history of almost half a century, it has only become a hot topic in the science community and the general public in the most recent decade. As these technologies are gradually recognized by the robotic community, more and more scientists and engineers look to contribute to the field. This is reflected by the everincreasing number of laboratories, international collaborations, emerging publications, soft robot-related societies and organizations, special session in all kinds of international conferences, professional events, and activities. Although the soft robotics field is still in its infancy, a number of review articles have been published to summarize the achievements, analyze the techniques, and discuss the challenges and prospects for the future. These reviews were organized in terms of technical contents but we would like to present a different perspective by using bibliometric analysis to show the historical map and overall view of the soft robotics research field.
Bibliometric analysis is quite effective for analyzing scientific publications to map the historical development of the target topic, find the hotspots, highlight the distribution layout of active researching countries, institutions, authors, and their cooperation relations, as well as top journals for publications, leading influence articles, and research trends. It has been adopted in a variety of disciplines, such as chemistry, 29 economics, 30 computing, 31 management, 32,33 education, 34 medicine, 35 energy, 36,37 and robotics. 38 Yet, to our knowledge, this is the first bibliometric analysis to assess the soft robotics research field. Our goal is to provide a general overview on this research area by revealing the following aspects: (1) historical map of the topic; (2) the main contributors: countries, institutes, research groups, authors, and leading research areas; (3) cooperation patterns between countries, institutes, and authors; (4) the most productive journals; (5) top articles with highest citation number; and (6) research interests and perspectives.

Methodology and Data Source
The analysis is based on the publications related to ''soft robot'' published from 1985 to 2017. Literature were retrieved through the Science Citation Index-Expanded and Social Science Citation Index on August 17, 2017, with search formula of ''Artificial muscle*'' or ''Pneumatic muscle* actuat*'' or ''continuous robot*'' or ''redundant robot*'' or ''soft robot*'' or ''soft wearable robot*'' or ''Bio inspired Robot*'' or ''Bioinspired Robot*'' or ''soft bodied robot*'' or ''bio soft robot*'' or ''biomimetic* robot*''or ''biological* inspire* robot*'' or biorobotic* or microrobotic* or ''bio robot*'' or bioactuat* or ''redundant actuat*,'' defining the document type as article and review in the field of topic. As a result, 1495 articles were collected from InCites data set including Web of Science (WOS) content indexed through May 31, 2017. Articles originated from England, Scotland, Northern Ireland, and Wales were grouped under the United Kingdom heading. Keyword and international cooperation were analyzed by Thomson Data Analyzer. The impact factor (IF) for each journal was determined according to the report of 2016 Journal Citation Reports. Since the WOS ''topic'' searching was applied to the title, abstract, and keyword fields defining the document type as article and review, some other related publications may not be covered.

Global contribution and leading countries
Although literature retrieving covers the time span from 1985 to 2017, articles concerning soft robots were first published in 1990. From then on, 70 countries have contributed to the soft robotics research field with 1495 publications, in which 37 are Essential Science Indicators (ESI) high cited articles and 4 are ESI hot articles.
The term ''soft robotics'' was initially used to represent rigid robot that had compliant joints and variable stiffness. 4 Then soft robots were distinguished from traditional robots and soft robotics became a new multidisciplinary field involving soft material-based structure with compliance and deformability in the interaction with environment. 5 From 2008, ''soft robotics'' was widely adopted as a keyword in scientific articles, especially after 2012, shown in Figure 1. The emerging trend of publications with the keyword ''soft robotics'' is consistent with the polyline shown in Figure 2. Although the first articles related to soft robotics emerged in 1990, the number of total publications per year was relatively stable ranging from 7 to 27 during 1990-2007, suggesting that this domain was not so attractive to scientists and engineers at that time.
But just 1 year later (2008), at the beginning of OCTOPUS IP, 39 the Large-Scale Integrating Project funded by the European Commission under the 7th Framework Programme, we can see publications increased to 41. The trend continued with increments of several articles until 2012, when the number of publications rose 66% to 101 articles on a year-onyear basis. Since then, the rising rate of yearly publications has been relatively stable, with high percentages of 17%, 30%, 35%, and 27% in 2013, 2014, 2015, and 2016, respectively. The active interests and intensive efforts from research and engineering communities in the past few years has led to the enormous growth of this field, which is supported by a large increase in the number of publications. 40 The total number of articles published in the most recent 5 years is more than three times of that on this topic in the first 18 years since 1990. In the first 5 months of 2017, there already exist 191 articles in the soft robotics field, which will certainly contribute to another bigger increase this year than the 260 articles in 2016. Table 1 shows the top 20 countries in terms of the number of publications related to the field of soft robotics. The United States is the most productive country with a total of 478 articles since 1990, followed by China (230 articles) and Italy (149 articles). Although we cannot attribute this productivity to particular causes, these countries are the focus of several new funding programs such as the DARPA ChemBots program in the United States in 2008, the major national research funding initiative Tri-Co Robot in China from 2016, and the OCTOPUS IP at the BioRobotics Institute of Scuola Superiore Sant'Anna 41 in Italy. The BioRobotics Institute is also the primary host of a Coordination Action for Soft Robotics funded by the European Commission under the Future and Emerging Technologies-FET-Open Scheme that hosted a series of activities and events. 40 Of the top 20 country's publications, a large number (>28%) are international articles, especially for the Netherlands (78.95%) and Germany (65%). This implies that soft robotics has attracted worldwide scientists and engineers to exchange ideas and cooperate with each other. Another observation is that despite the high number of publications from China (second with 230 articles), the average citations per publication (ACPP) is relatively low, only 7.7%. It is unclear whether this reflects a language barrier, a bias in accessing different publications, or the scope and quality of the research itself. Figure 3 shows the collaborative relationship of the top 20 productive countries. The size of nodes is proportional to the total number of articles of each country. The lines represent collaboration between countries, the thickness of which indicates the intensity of cooperation. The United States is the most active country that collaborated with 50 countries, especially with China, Italy, Germany, South Korea, and Japan. Germany lists on the second place, followed by Italy and the United Kingdom. One of the possible reasons might be the Europe visa policy that makes the European research institutes easy to recruit researchers from other countries within Europe. Another reason relies on the European research council that provides lots of cooperation opportunities for researchers from different European countries. Table 2 shows the top 20 productive institutions in soft robotics research along with their total number of publications, citations, and h-index. Apparently, most of them are from top 10 productive countries. Harvard University leads the list with the most publications followed by Scuola Superiore Sant'Anna and Chinese Academy of Sciences. As for the ACPP, Harvard University and Massachusetts Institute  Table 2 (22.13). However, the four Chinese institutions in the top 20 list have relatively low ACPP results, <10. Compared with the influence of the American and Italian research institutions, the Chinese counterparts need more efforts to improve their research work and global influence in this field to match with its position in publication list.

Contribution of leading research areas
As is well known, soft robotics is a quite new developing multidisciplinary field, 42 which is also supported by articles distributed in 91 WOS research areas. Table 3 illustrates the top 20 WOS research areas ranked by the number of articles related to soft robotics. There is no doubt that ''Robotics'' dominates the research area list with 557 articles, followed by ''Automation and Control Systems,'' ''Engineering, Electrical and Electronic,'' ''Engineering, Mechanical,'' ''Materials Science, Multidisciplinary,'' and ''Computer Science, Artificial Intelligence,'' which are the main scientific areas that put special emphasis on soft robotics. ''Physics, Condensed Matter,'' ''Chemistry, Multidisciplinary,'' and ''Chemistry, Physical'' lead the list of ACPP, with ACPP of 27.33, 25.75, and 24.73, respectively. These three areas are closely related with the current hot topic in soft robotics: smart material, even biohybrid material. The high ACPP in these areas verifies the widely accepted opinion that material is the key for soft robot development. 2,[43][44][45] Leading journals in terms of number of publications in soft robotics research The 1495 articles related to soft robotics during 1990-2017 were published in 357 journals. As listed in Table 4, IEEE Transactions on Robotics takes the leading position with 64 articles, followed by Bioinspiration & Biomimetics (62), IEEE-ASME Transactions on Mechatronics, (57) and Soft Robotics (53). The aforementioned four journals share 15.79% of the total publications, and the top 20 journals listed in Table 4 have produced 656 articles with the share of 43.88%. And all the remaining journals contribute with shares less than 1% each. In terms of IF, Soft Robotics has the highest value of 8.649 except the two material-related journals Advanced Materials and Advanced Functional Materials. Because of its overwhelming high IF, Soft Robotics holds the top position in journals categorized with ''ROBOTICS'' in  Table 5.
To show the historical map of soft robotics-related publications in journals, we employ the bubble chart of top 20 productivity journals by year, shown in Another journal, Science Robotics launched at the end of 2016, 46 is not included in this analysis. However, in the published 9 issues, 13 out of 41 total articles are soft robotics related. The 31.71% of soft robot-focused articles indicate the deep interest and close attention of the journal in this field. Table 6 shows the top 10 most productive authors based on the number of publications. Whitesides group leads the list  with the total number of publications of 18 followed by Wood (13), Laschi (12), Cianchetti (12), Yu (12), and Walsh (12). For the ACPP, Whitesides group ranks the first with number of 69.33, followed by Wood (34.85), and Laschi (32.25). Whitesides group also achieves the highest h-index of 14, followed by Wood (11) and Laschi (6). Interestingly, Prof. Whitesides laboratory is based in a Chemistry Department that enables him to develop soft robots from the viewpoint of chemistry and materials. 45 Table 7 illustrates the top five authors with high total citation, all of whom are from Whitesides research group. All these top 10 productive authors are from the top 3 productive countries, indicating that there are certain concentrating groups in this field, such as the Whitesides research group at Harvard University and the Octopus research group in Europe (mainly at Scuola Superiore Sant'Anna, Italy). Nevertheless, the share of articles of the top 10 authors in total publications is only 7.88%, which means that a large number of researchers are working in this field and make contributions in the total 1495 publications. The vast population in this research community will certainly yield extraordinary progress and diverse achievements in the near future.

Analysis of the most cited articles
Although the citation impact of an article will be influenced by many factors, 47 it is still a widely accepted index for evaluating scientific articles. We list the top 10 most cited publications for analysis, shown in Table 8. The most highly cited article is ''Self-organization, embodiment, and biologically inspired robotics'' published in Science by Pfeifer et al. 48 It leads the list of total citations (356). Whereas ''Multigait soft robot'' 49 and ''An electrically and mechanically self-healing composite with pressure and flexion-sensitive properties for electronic skin applications'' 50 take the second and third places according to their total citations (353 and 330) with quite high annual citations (50.4 and 55), respectively. Moreover, it should be noted that the recently published article ''Design, fabrication and control of soft robots'' (2015) 42 in Nature gains the highest total citation per year (69), which to some extent reveals the wide concern on this field. Among these top 10 articles, 6 were published in top journals such as Science, 48 Nature, 42,50,51 and IEEE transactions. 52,53 Extraordinarily, eight are from institutes in the  Again, the Whitesides group shows its position and influence in soft robotics by contributing 2 articles 45,49 among the 10 most highly cited publications. Moreover, ''Design, fabrication and control of soft robots'' (2015) 42 holds the first position of the ESI hot article, followed by ''Stretchable, skinmountable, and wearable strain sensors and their potential applications: a review'' (2016), 55 ''Soft robotic glove for combined assistance and at-home rehabilitation,'' (2015) 56 and ''Phototactic guidance of a tissue-engineered soft-robotic ray'' (2016). 57

Research interests and perspectives
As listed in Table 9, the first review on soft robotics was published in 1999. 58 There was no other review for a decade before soft robotics attracted wide interests in 2008. After that, there were reviews on soft robotics each year. Especially after 2014, there were at least four reviews on this topic, which again implies that soft robotics is a hot spot in the robotic community. The multidisciplinary nature of soft robotics and the variety of authors' professional backgrounds result in a diversity of contents, analysis perspectives, and arguments in their publications. Despite the different perspectives, all the reviews listed in Table 9 were conceived from the technical standpoints, such as actuator, fabrication, control, material, sensing, simulation, bionics, stiffness, modeling, power, motion, and application, as shown in Table 10.
Most of the reviews cover the latest work in this field around the world and summarize technical aspects as listed in Table 10. The reviewers always conclude with comments on the current problems and prospects for future work. These summaries, comments, and suggestions on techniques are important for fellow researchers and potential followers in this field. Thus, readers are encouraged to refer to the review documents listed in Table 9. In addition, these articles include discussions on new research directions and ideas that  Seoul National University are expected to guide and stimulate creative work. Smart materials were mentioned in most reviews (referring to Table 10) and deeply discussed in most documents. 5,43,59,60 Walker 59 and Iida 28 believe that collaboration between materials science and engineering will benefit soft robotics rather than isolated work. Bionics is also seen as a very inspiring source for soft robots, yet most articles caution that learning from biological form 61 and motions 59,62,63 is not the whole story. Biology-originated techniques or ideas, such as tissue engineering, 2,64 camouflage, 65,66 self-cleaning and self-healing, 5,65 and even growth, 5 are expected be integrated into the field to create life-like (or even living) soft robots. Another grand challenge for soft robot development is rapid virtual prototype techniques such as 3D printing, 59,67 which require special modeling and simulation tools, 42,44,62 totally different from those for rigid robots. 68 Furthermore, some new advanced ideas were proposed, such as mechanical intelligence 64 and task distribution 62 for soft robot construction, modeling, and control. Pfeifer et al. 62 and Iida and Laschi 44 have advocated morphological com-putation and embodiment control for soft robot, which is also agreed by Rus and Tolley. 42 Like conventional rigid robots, soft robot research started from social driving forces 63 and it is expected to develop to commercial products. 2,42,62 This will involve solving significant challenges of cost and safety, 28,64 power supply, 28,43,62 harvesting, 65,69 and user and operator interfaces. 59 Despite all of these aspects, the development of useful and sustainable soft robots will require advances toward a common fundamental theory that forms an infrastructural framework. This theoretical work requires both a deep knowledge of mathematics and a thorough understanding of soft robots. This is a grand challenge for the new discipline and it requires scientists and engineers with different scientific backgrounds, such as robotics, mechanical engineering, materials science, computer science, controls, chemistry, physics, biology, and mathematics. Furthermore, they are expected to cooperate intensively with each other to make substantive achievement in theoretical research.   Although soft robotics is a relatively new field, there is no doubt that it is a rapidly growing topic in robotics. This is supported by the emerging publications, new related journals, and general public interest. The field is widely spreading and growing rapidly with institutions in the United States, Europe, and Asia contributing most of the original research, developing the infrastructure, and consolidating the academic community.
The growth of soft robotics requires new developments in complex structures, sensing, control, and power supplies that increasingly rely on advancing materials science and the corresponding manufacturing techniques. This is reflected in the number and prominence of articles in this field, discussing smart material applications and the appearance of materials science journals in the top list of soft robotics publications.
In reference to the journals, IEEE Transactions on Robotics ranks first among the top 20, whereas Soft Robotics holds the top position in journals categorized with ''RO-BOTICS'' in the latest two consecutive years.
Whitesides, Wood, and Laschi are the top three most productive authors assessed by the ACPP and with highest h-index ranks. Furthermore, Whitesides research group occupies all the positions of top five authors with high total citations, which implies their powerful influence in the global soft robotics community. In addition to the top 10 authors, a large number of researchers are working in this field and they contributed 92.12% of the publications.
An analysis of reviews on soft robotics suggests that the following aspects are most attractive to researchers: actuators, fabrication, control, materials, sensing, simulation, bionics, stiffness, modeling, power, motion, and applications. Furthermore, multidisciplinary areas such as smart material, bionics, morphological computation, and embodiment control are expected to be major contributions to the field in the future. Some reviews discussed obstacles in the way of commercialization, but the development of a common fundamental theory for robot design and control seems to be a vital need for soft robotics to be an independent discipline.