1 Introduction

Physical activity has a wide range of positive impacts on humans, ranging from physiological effects, such as strength and stamina or lower obesity rates (Stodden et al. 2008), to psychological effects like higher well-being (Fox 1999) and lower consumption of alcohol (López Villalba et al. 2016; Malm et al. 2019). Furthermore, active individuals remain more active even in old age (Kjønniksen et al. 2009) and exhibit a higher knowledge of nutrition, exercise, and health (Khan et al. 2012). Despite various benefits, many individuals, especially high school students, perceive a lack of motivation (Papaioannou 1997; Deelen et al. 2018). One factor that contributes to the athlete’s motivation is the presence of competition (Frederick-Recascino and Schuster-Smith 2003). However, physical competition is not always feasible, for example due to long travel times, reduced mobility, or weather conditions. One promising solution to all these issues is the usage of virtual reality (VR) sports.

Virtual reality is a technology characterized through four main attributes: immersion, interactivity, presence, and togetherness (Walsh and Pawlowski 2002). Immersion is the extent to which users can fade out the real world and are stimulated by the virtual world (Witmer and Singer 1998), while interactivity is the modification of elements in this virtual world (Steuer 1992). Presence refers to the feeling of being present in another environment (Witmer and Singer 1998) which can be shared with others, called togetherness (Durlach and Slater 2000). These characteristics are usually achieved through wearing so called head-mounted displays (HMDs). Prominent examples are the HTC Vive or Meta Quest (Steam 2024). Virtual reality and augmented reality (AR) belong to the spectrum of extended reality (XR) technologies. However, in contrast to VR, AR enriches the physical world with virtual elements (Milgram et al. 1995). The most prominent application areas of VR are gaming, travelling, and watching movies or videos (Statista 2023). Nearly 30% of the study participants of a survey in Germany declared that they have used VR for sport activities and 19% of those even participated in virtual sport events (Statista 2023). These results indicate an overall high interest in the combination of virtual reality and sports.

Existing studies in the field of VR sports highlight several advantages: increased fun, enjoyment and satisfaction (Hassandra et al. 2021; Caserman et al. 2022; Polechoński et al. 2024), the improvement of skill quality (Michalski et al. 2019; Mologne et al. 2022), and of motivation (Pu and Yang 2022; Keller et al. 2022; Gulec et al. 2023). Especially for individuals with reduced mobility and elderly persons, VR environments can be adjusted to meet their special needs (Kwan et al. 2021; Høeg et al. 2023; Yu et al. 2023). VR sport can also reduce the perceived exhaustion and increase endurance (McMahon et al. 2020; McDonough et al. 2020; Saiz-Gonzalez et al. 2023). Similarly, for children, whose parents often call for more physical activity, VR sports can be a great opportunity (McMichael et al. 2020). However, some studies could not measure an enhancement of the training effect (Witte et al. 2022; Oagaz et al. 2022a; Rutkowski et al. 2022) or observed negative side effects, such as motion sickness (Rodrigues et al. 2020; Kwan et al. 2021). Some study participants also claimed that they were missing social aspects in the tested VR applications (Høeg et al. 2023; Karatas et al. 2023).

Psychological models can help us to understand the potential benefits of VR sports. As VR has the potential to satisfy psychological needs, it raises the question if VR sports applications are designed to maximize intrinsic motivation, engagement, and the learning of new skills in physical activities. According to the cognitive load theory, learning and skill mastery are influenced by intrinsic cognitive load related to the complexity of the task, extrinsic cognitive load depending on the presentation of information, and germane cognitive load, which involves the processing and integration of new information (Sweller et al. 2011). VR applications can optimize cognitive load by presenting physical exercises in an interactive manner (Haryana et al. 2022). The situated learning theory says that authentic contexts can enhance learning by providing experiences that mirror real-life settings and include social interaction (Lave and Wenger 1991). VR offers new possibilities to interact with other users and is accessible from anywhere at any time. Hence, it could encourage users to form social connections with others from home. On the other hand, it has the capabilities to simulate realistic real-world scenarios (Scavarelli et al. 2021). Thanks to these benefits, VR not only offers the benefits of practicing in the real world, but also the opportunity to support real-world training in social, accessible, and safe environments. The self-determination theory emphasizes the role of intrinsic motivation in our behavior and learning. According to the self-determination theory, the fulfillment of the basic psychological needs for autonomy, competence, and relatedness is essential for intrinsic motivation and optimal well-being (Ryan and Deci 2000). The immersive nature of VR applications can enhance the intrinsic motivation to engage in sport and fitness activities by presenting them as enjoyable and rewarding experiences (Huang et al. 2019; Reer et al. 2022).

The rising interest in VR sports is seen in bibliometric analyses that highlight the rising number of annual publications over the last two decades (Denche-Zamorano et al. 2023). It is also shown that the most important topic within the context of VR fitness is rehabilitation. Although the taxonomy approach in the field of information systems is not widespread, some taxonomies already exist for different VR fields. Examples include VR applications for the treatment of anxiety disorders (Korbel et al. 2021), the use of VR in education (Motejlek and Alpay 2021), or VR for cultural heritage practices (Chong et al. 2022). However, we are not aware of such research for VR sports applications. Furthermore, first literature reviews summarize knowledge on VR sports, for example for non-immersive VR (Akbaş et al. 2019), a small number of VR sports studies (Richlan et al. 2023), or visual anticipation for sports performance improvement through VR technology (Heilmann and Witte 2021). However, there is a lack of systematic reviews of VR sports environments, which is why a new literature review, focusing on the applications themselves, is necessary as a foundation for the taxonomy. By creating a taxonomy of both scientific and commercial applications and comparing results, this study contributes to the research field of VR sports by structuring it and identifying trends and research gaps. Further, we evaluate whether current VR sports applications are suitable to learn new skills based on established research theories. With our taxonomy, we want to close the knowledge gap regarding a structured overview of VR sports applications by answering the following research questions:

  1. 1.

    Which similarities and differences exist between research and commercial VR sports applications and which research gaps derive from that comparison?

  2. 2.

    Are current VR sports apps designed to effectively support learning?

The scope of our article includes VR sports applications running on HMDs, while excluding other forms of virtual reality, like Cave Automatic Virtual Environment (CAVE) or phone-based VR, with the intent of ensuring a high degree of immersion. In contrast, the differentiation between sporting applications and non-sporting applications is less clear-cut. One definition of sports includes all human activities that involve physical exertion, skill, and competition, governed by formal rules and organizations (Commonwealth of Australia 2011). However, especially in the field of virtual reality, where nearly all apps include human movements, it is challenging to draw a clear line between exergames—applications that combine physical exercise with game elements—and games without sports character. We define VR sports applications as applications that (1) can be assigned to a traditional sport, or (2) need to be executed while standing, or (3) can be executed while seated, but require strenuous movements. In edge cases, we discussed specific applications in a group of two researchers.

This paper starts by explaining the methods used in Sect. 2. Afterwards, we summarize literature in Sect. 3 and develop the taxonomy in Sect. 4. After discussing our results in Sect. 5, we conclude the article in Sect. 6.

2 Methods and material

A solid foundation of knowledge is required prior to developing a taxonomy, which is the reason for first conducting a systematic literature review following the PRISMA guidelines (Page et al. 2021) and then creating the taxonomy based on the guidelines by Nickerson et al. (2013).

2.1 Systematic literature review process

For the systematic literature review, we followed the PRISMA guidelines (Page et al. 2021) and focused on the most relevant items for our specific use case. The primary outcome of this research is the development of the taxonomy. As such, certain elements like pre-registration, which do not support this goal directly, were not conducted in this study. We queried two search interfaces: WebOfScience and Scopus. In both cases, we used the search string (“Virtual Reality” OR VR OR “Head-Mounted Display” OR HMD) AND (Sport OR Fitness) for Title/Abstract/Keywords. This search string ensures that all articles include both VR devices and sports aspects. To ensure a broad range of possible sport disciplines and prevent any bias before conducting the review, we deliberately kept the search string broad. We acknowledge that this approach may not capture every specific sport. However, our initial queries already returned a large number of studies and a subsequent forward–backward search allowed us to identify many additional studies for several sports disciplines.

Our search resulted in 3691 publications in total that were screened by two of the authors. As shown in Fig. 1, the literature screening process consisted of different phases. Before the screening, we removed duplicates from the dataset. Further, we conducted a pre-screening to remove non-English records, records without available full text, and retracted articles. From the remaining 2954 records, we excluded those whose document type differs from an article and all articles older than 2012, since the first commercial HMD, the Oculus Rift, was released that year. The remaining 1868 records were then assessed for eligibility. We applied five exclusion criteria in this step to exclude:

  1. 1.

    Articles whose content is out of scope,

  2. 2.

    Articles that include VR but have no focus on sport,

  3. 3.

    Articles that focus on sports but do not utilize VR,

  4. 4.

    Articles that either utilized non-immersive VR or did not specify the headset used, and

  5. 5.

    Articles that did not implement individual applications, since we want to compare research applications with apps on the market.

Fig. 1
figure 1

PRISMA diagram showing the literature selection process of the systematic review, leading to the final set of 59 publications

Full size image

Finally, we conducted a forward and backward search resulting in 8 additional articles. From the final set of 59 studies, we extracted relevant data in the next step, such as headset, controllers, sport discipline, or avatar. These categories were developed in an iterative process by analyzing the information provided in included publications. However, not all articles describe their application in detail. For some categories, a “not specified” label had to be used.

2.2 Application review proces

To find suitable applications, we relied on the most prominent VR app stores, namely the Meta Quest Store and the HTC Vive Store (VIVE Port). We queried both app stores with the search strings “sport” and “fitness”. We used a text-based query instead of relying on predefined sports categories as this returned several additional apps that were not included in the store-provided categories. The app selection process is shown in Fig. 2. First, we removed 124 duplicate applications. We manually screened the remaining 408 applications for eligibility by applying four exclusion criteria. More concretely, we removed:

  1. 1.

    Apps whose content is out of scope,

  2. 2.

    Apps that only extend others, i.e., game expansions and DLCs,

  3. 3.

    Gaming apps without a sports character, and

  4. 4.

    Meditation apps.

Fig. 2
figure 2

Selection process of VR sports applications published on VIVE Port and Meta Quest Store

Full size image

For a more consistent classification of the apps, two researchers screened all apps from the initial sample and labeled them. Differences in the classification were discussed for each application. The final 141 applications were used to create the taxonomy. Because of the large size of the data set, we decided to group them based on the implemented sport and choose a representative application for each group that was screened in detail. To choose adequate representatives, we decided to select apps that are as heterogeneous as possible to achieve a taxonomy representing a wide range of potential VR sports applications. We also paid attention to the amount of information available for the representatives. An overview of the representative application for each sport can be found in Table 1. With the applications, we intended to represent the most prominent commercially available sport categories for VR.

Table 1 Representative applications used for the taxonomy creation
Full size table

2.3 Taxonomy creation process

A taxonomy is a categorization of objects within a domain based on shared characteristics. The taxonomy creation process is based on the method of Nickerson et al. (2013). They define it as a set of one or more dimensions with several characteristics each. For example, one of the dimensions of our taxonomy is the user’s avatar. In the first iteration, based on the literature review, we identified broad characteristics, such as whether the player acts as the athlete or referee. However, when analyzing commercial applications, we noticed that some apps relied on animalistic or torso-only avatars, requiring a subsequent refinement of this dimension to include further characteristics, a finer characterization and to cover all available applications. This adjustment continued until all applications were adequately classified within the taxonomy.

The taxonomy creation process consists of three phases. The first step is to determine the meta-characteristics. Meta-characteristics are overarching properties and define the basis for the taxonomy. They are derived from the purpose of the taxonomy. Relevant dimensions for categorizing specific objects should then refine the meta-characteristics. Secondly, ending conditions are defined that determine when the taxonomy creation process is complete. There are two types of ending conditions: objective ending conditions, e.g., when all objects were analyzed, and subjective ending conditions, e.g., the explainability of a taxonomy. We adopted all ending conditions, objective and subjective, from Nickerson et al. (2013).

The final step of the taxonomy creation process includes several iterations until all ending conditions are fulfilled, e.g., until all objects can be categorized adequately with the taxonomy. Performing multiple iterations allows for a more granular division of objects into sub-sets that can be examined more thoroughly compared to a single iteration. Iterations can be either empirical-to-conceptual or conceptual-to-empirical. While the empirical-to-conceptual approach starts with a subset of empirical objects that are moved to dimensions and characteristics, the conceptual-to-empirical approach begins with conceptualizing characteristics and dimensions before specific empirical objects are examined. Hence, the first iteration of our taxonomy follows the conceptual-to-empirical approach, starting with dimensions and characteristics derived from theory, i.e., the literature review from Sect. 3. The following iterations were empirical-to-conceptual, based on commercial applications. We added new characteristics in each iteration if (1) the regarded VR sports applications do not fit into the current taxonomy or (2) new dimensions appear that are needed to distinguish between different properties of the VR sports apps.

Finally, both approaches, conceptual and empirical, converge into the last step where the ending conditions are checked. If all ending conditions are fulfilled, the taxonomy creation process is finished. Otherwise, a new iteration needs to be performed (Nickerson et al. 2013). While the objective ending conditions are easy to verify, we relied on several methods to verify all subjective ending conditions. On the one hand, we often relied on characteristics summarizing different cases, e.g., the characteristic “training facility” to summarize “boxing ring”, “tennis court”, and “golf court”, and limited the number of dimensions to 15 to maintain conciseness and avoid overwhelm. The grouping helped us keep the taxonomy concise, while retaining comprehensiveness by ensuring that all relevant expressions of a dimension are still covered by the taxonomy. On the other hand, we ensured robustness by testing the taxonomy with further VR apps that were not in the set of representative applications. The comprehensiveness and explainability were further ensured by two expert interviews. Finally, the extendibility of the taxonomy was verified with three VR sports applications that were not part of the original analysis.

Unlike in the original definition of a taxonomy by Nickerson et al. (2013), we decided to allow the classification of VR sports applications into several characteristics of one dimension, if needed. This allows us to waive the listing of all combinatorial possibilities if applications for instance implement several sceneries. Since our literature review delivers first insights into possible VR sports applications, we derived the first version of the taxonomy from the categories and features from the literature review. We analyzed the commercial applications in the following iterations to refine the taxonomy.

3 Systematic literature review

In this section we preform a systematic literature review, based on the PRISMA guidelines (Page et al. 2021). The goal is to derive a foundation of knowledge to aid the creation of the taxonomy.

3.1 Sport categories

We encountered a wide range of sports in the selected articles, as shown in Table 2. One of the most prominent categories of the regarded studies were exergames, i.e., active applications characterized by playfulness that are not directly related to any traditional sport discipline. Examples of exergame applications in the literature include a game where users need to hit viruses (Yang et al. 2022), a game where players need to squat to avoid obstacles while standing on a moving train (Ou et al. 2020), or punching spiders and monsters attacking the user (Graf et al. 2023). The most prominent conventional sport in the literature is cycling, where users typically sit on ergometers with an HMD that shows an environment that moves along during the ride. Other frequently examined sports are soccer, either from the perspective of a player or a goalkeeper, weight training, and boxing. However, we also observed a wide range of sports with only one or two publications each.

Table 2 Frequency of sport categories in the analyzed literature
Full size table

3.2 Hardware

The category of hardware splits into technical input, output, and possibly additional external devices. Regarding the design of any VR application, it is crucial to work with the limitations of current headsets. Headsets wired to computers benefit from the better performance but reduce the user’s mobility, while the opposite applies to their wireless counterparts that provide flexibility of movement but are generally less performant. In addition to headset types, the usage of controllers also presents a difference between different studies. Depending on the type of sport, holding a controller can seem unnatural for intuitive motions, adding another layer of complexity to VR sports applications. Designers of VR environments need to be aware of such limitations. An overview of the headsets used can be found in Table 3. The most prominent headsets in the considered studies are the HTC Vive and Oculus Rift. Both are wired headsets that need to be connected to computers. However, according to a statistic by Steam, the market share of those headsets is decreasing while wireless models, such as the Meta Quest are gaining in popularity (Steam 2024). Quest headsets are the third most frequently encountered option in the examined literature. Only a small number of studies used different devices.

Table 3 Headsets in the regarded literature
Full size table

Most studies relied on controllers for user input, as shown in Table 4. They were either handheld or attached to an external device, such as a tennis racket (Gaugne et al. 2022) or on an ice hockey stick (Polikanova et al. 2022). Despite a large number of studies not specifying the input device, it can safely be assumed that controllers were used due to the inability of some HMDs to track hand movements. Finally, a small minority of studies specified that they directly tracked the user’s hands.

Table 4 Input devices in the reviewed literature
Full size table

As Table 5 shows, beside controllers, some of the studies included additional devices that were typically specific to the sport. All studies examining VR cycling include either an ergometer or at least pedals for incapacitated individuals that need to sit in a chair. The Xbox Kinect, a device that can track the player’s whole body, was also used in some studies. The movements are then displayed in the VR environments to mirror the current body position. The studies on (table) tennis included rackets to make the experiences more realistic, while other studies included basketballs, bats, or weights.

Table 5 External devices other than controllers used in the literature
Full size table

3.3 Target group

The target groups of the studies that we considered are shown in Table 6. Most of the studies did not specifically mention a target group and are therefore assumed to have heterogeneous samples. Studies that did focus on a specific target group were most commonly tested with (professional) athletes, e.g., football applications for football players or goal keepers. However, some experiments were conducted with beginners in specific sports with the aim to teach them basic movements rather than refining their technique. While the majority of publications focus on healthy participants or did not specify a focus, some studies explicitly target individuals with reduced mobility, COVID-19 patients, or overweight individuals. Only a small number of publications directly targets children or elderly people.

Table 6 Target groups of the considered applications
Full size table

3.4 Characters and avatars

As a representation of the player, some applications implemented avatars. All different kinds of avatars are listed in Table 7. Most studies that used avatars designed them in a way that is typical for the sport under investigation, e.g., football players, ice hockey players, or karate athletes. However, some studies relied on minimalistic avatars, meaning only in the shape of a human without complex textures. Often these avatars are monochrome, e.g., blue or red, depending on the team in the game (Vu et al. 2022). While some studies did not specify how their avatars look like, one study even implemented a referee avatar.

Table 7 Player avatars in the considered applications
Full size table

In addition to the player’s avatar, many studies also included other characters, as shown in Table 8. Most prominent are the instructors or assistants, i.e., a person guiding the execution of specific movements. These instructors are usually pre-recorded. Further, some studies implemented one or several opponents. However, these opponents were always controlled by a computer. The same number of research applications also implemented teammates. Spectators and referees on the other hand were rather rare in the literature.

Table 8 Other avatars or characters in the considered applications
Full size table

3.5 Application design

The applications in the literature exhibit various application designs, ranging from gameplay variations to the implemented scenery. An overview of implemented gameplay variations is given in Table 9. The levels of the applications in the literature target mostly different tasks, e.g., emergency situations, hazardous environments and high-risk job tasks without being exposed to any real danger or injuries, e.g., in the case of VR skiing (Li et al. 2023). Different free kick scenarios are another example in a goalkeeping simulation (Valkanidis et al. 2020). These applications are summarized in the category of level diversity. Some applications designed levels with increasing difficulty, called difficulty scaling. For example, a tennis ball was served from increasingly large distances from the player (Gaugne et al. 2022).

Table 9 Gameplay variation in the considered applications
Full size table

To train different sports, the research applications took place in front of different sceneries, as presented in Table 10. The most prominent sceneries in the applications are stadiums with spectator stands surrounding the area. Gym environments are similarly popular. Here the background consists of either a room with fitness equipment or a larger sports hall. However, some of the studies also implemented natural environments or a range of different environments. For example, exergame applications often applied different environments to their games, depending on their context (Ou et al. 2020; Wu et al. 2023). Other environments we encountered frequently are sport courts, e.g., tennis courts (Le Noury et al. 2021; Gaugne et al. 2022) or a golf court (Harris 2019), and natural environments, like lakes (Cao et al. 2021) or forests (Liang et al. 2018; Ju et al. 2023). Environments, such as streets, empty rooms, or boxing rings were rather uncommon.

Table 10 Gameplay scenery in the considered applications
Full size table

4 Taxonomy

The taxonomy creation consists of the taxonomy preparation—determining meta-characteristics and ending conditions—and the iterations themselves until all ending conditions are fulfilled. While we derived the initial taxonomy from the literature review, we used the remaining iterations to analyze commercial applications we found in the app stores. We present the resulting taxonomy at the end of this section.

4.1 Taxonomy preparation

For the first step of the taxonomy preparation, determining the meta-characteristics for the taxonomy, we choose “virtual reality sports applications available in the most popular virtual reality app stores”. Further, we apply the ending conditions by Nickerson et al. (2013). The whole taxonomy creation process is shown in Fig. 3.

Fig. 3
figure 3

The first six iterations of the taxonomy creation process that develops from concepts derived from literature to concepts derived from commercial VR applications

Full size image

4.2 Iterations

4.2.1 Iteration 1

In the first iteration, we follow a conceptual-to-empirical approach, with the literature review serving as the foundation. We directly derive the dimensions from the literature categories in Sect. 3 and convert their features into characteristics. As we have added new dimensions and characteristics in this iteration, another iteration is necessary.

4.2.2 Iteration 2

In this iteration, we adopt an empirical-to-conceptual approach, deriving dimensions and characteristics from VR applications focusing on team-oriented sports, as these show great popularity. Traditional team-oriented disciplines have a wide appeal and are well-established in both physical and virtual formats. We examined five different VR sports applications in our database: P1 (WIN Reality Baseball), P2 (ib Cricket), P3 (Gym Glass—Basketball VR), P4 (CleanSheet Soccer), and P5 (2MD: VR Football). Since bat-and-ball sports, as well as ball sports without a bat, encompass a wide range of possible sports, we decided to group similar sports into broader characteristics instead of listing each sport individually as separate characteristics. All products are designed for the Meta Quest headset. However, two applications are also compatible with HTC Vive and Valve Index (P2 and P5). We decided to limit the “hardware” dimension to the three most prominent HMDs as apps are usually created for the most popular devices. However, we introduced an additional characteristic for other headsets. Furthermore, all five products use controllers for input and the user avatar imitates a professional athlete. In terms of gameplay, all applications involve interaction with other avatars, such as teammates and opponents, and are set in sports venues. We decided to summarize the various venues specific to the respective sport as a single characteristic. Furthermore, P1, P2, P3, and P4 offer multiple diverse levels, while P2 additionally offers narrative progression which we added as a new characteristic. We also retrieved key characteristics from the app stores. We noticed that all applications support multiplayer, allowing users to engage with other players. The games can be played in various positions—either seated (P1, P5) or standing (P1, P2, P3, P4). The applications P1, P2, and P3 offer in-app purchases, allowing users to buy additional content or features within the game. P2 and P5, on the other hand, do not offer such possibilities. This information requires new dimensions for the multiplayer functionality, the gaming positions to play the applications, and for the possibility to purchase in-app items. As we have identified specific characteristics of these five products in this iteration and added new dimensions, we need to perform another iteration to further refine and expand our taxonomy.

4.2.3 Iteration 3

For the third iteration, we continue with the empirical-to-conceptual approach, examining various sport applications that, contrary to team oriented disciplines, are played individually. They provide a contrast with respect to their unique environments, intensity, and activities. We examined four distinct VR sports applications: P6 (The Thrill of Fight), P7 (Golf +), P8 (VR Regatta), and P9 (Carve Snowboard). Since these applications represent different sports, i.e., boxing, golf, sailing, and skiing respectively, we need to update the “sport discipline” dimension. All applications, except for P6 and P9, support multiplayer, allowing users to engage with other players. In terms of in-app purchases, P7 and P8 offer new content, while P9 does not specify the type of in-app purchases. As previous applications offer cosmetics while P7 and P8 offer new content only, this observation led to the refinement of the “in-app purchases” dimension, splitting it into “cosmetics”, “content”, and “not specified” to differentiate the type of purchases properly. As we have identified specific characteristics of these applications in this iteration, we need to perform another iteration to further refine and expand our taxonomy.

4.2.4 Iteration 4

For the fourth iteration, we continue with the empirical-to-conceptual approach, examining racket sports and rhythm games. These add another layer of interaction and skill requirements, both using intense and immersive controller movements and showcasing the gameplay diversity enabled by the VR technology. We examined five additional VR sports applications: P10 (First Person Tennis), P11 (Eleven Table Tennis), P12 (Racket: Club), P13 (Dance Dash), and P14 (Synth Riders). These applications show the need for further refinement in our taxonomy. Specifically, we observe that P10, P11, and P12 are better classified under a shared characteristic ("racket sport") within the “sport discipline” dimension. Furthermore, we identify P13 and P14 as rhythm games, necessitating the addition of a new characteristic ("rhythm game") under the “sport discipline” dimension. This addition accounts for games that involve music and dancing as their central element of gameplay. In terms of external devices, P13 introduces a new method of control by using track straps. These are strapped to one of the player’s legs and lead to the addition of a new characteristic. While both P13 and P14, being less traditional sports games, are played in futuristic looking environments, P11 further offers an urban scenery. Finally, we noticed that P14 offers a multiplayer mode where two players compete on separate maps, with the player with the highest score being declared as the winner. This indirect competition differs from the direct competition observed in games like P10, P11, and P12 and previous team sports where players directly interact with each other. Therefore, we refine the “multiplayer” dimension to include specific types of competition. We introduce “direct competition” and “indirect competition”, in cases where games provide no direct competition but leaderboards to compare their high score with other players around the world, as new characteristics, as well as “single-player only”. As we have identified specific characteristics of these five products in this iteration, we need to perform another iteration to further refine and expand our taxonomy.

4.2.5 Iteration 5

For the fifth iteration, we continue with the empirical-to-conceptual approach, examining VR applications usually categorized as exergames. Exergames focus on physical fitness and exercise, blending elements of gaming with workout routines. This combination is unique to virtual applications, and they highlight how VR can be used not only for entertainment but also for promoting health and fitness. We examined four distinct VR sports applications: P15 (LES MILLS BODYCOMBAT), P16 (Holofit by Holodia), P17 (The Climb 1/2), and finally P18 (Xponential+). P15 and P16 are identified as exergames, P16 additionally contains cycling, and P17 represents climbing, reinforcing the existing characteristics within the “sport discipline” dimension, and finally P18 represents an application for different mind–body practices such as Pilates. P16 highlights the need for further refinement in our taxonomy. Instead of relying on controllers, P16 can also utilize hand tracking, introducing a new method of interaction. This observation leads to the creation of a new dimension, "tracking", with characteristics including “roomscale”, “hand tracking”, and “none”. In terms of payment models, most of the previously examined applications followed a pay-to-play model, with one exception that was free-to-play (P3). However, P16 and P18 introduce a subscription model, adding a new type of payment model to our taxonomy. This observation leads to the introduction of a new dimension, the "payment model", with characteristics including "pay-to-play", "free-to-play", and "subscription". As we have identified specific characteristics of these three products in this iteration, we need to perform another iteration to further refine and expand our taxonomy.

4.2.6 Iteration 6

Although all representative applications were examined in the last iterations, our database contains many more applications that have not been screened in detail yet and therefore need to be classified in our taxonomy. Thus, we stick with the empirical-to-conceptual approach. During the classification phase of the remaining applications, we recognized some dimensions that require further characteristics. We added a “miscellaneous” characteristic for the “sport discipline”, which includes apps with no clear focus on a specific discipline. Further, “animal”, “casual”, and “torso-only” characteristics were added for the “avatar” dimension to better classify the different types of avatars used in the remaining applications. Some of the remaining applications did not specify their gameplay variation and implemented fantasy/sci-fi environments. One of the research applications uses the gaming position lying, leading to a new characteristic. Further, we could identify applications that allow for song package purchases inside the apps. Moreover, we summarized the two dimensions “external devices” and “target group” where we created summarizing categories for their characteristics. Finally, we added the dimension “gamification” to the taxonomy to distinguish between applications using playful elements and applications that have a rather serious character. Since new characteristics and one new dimension were added, we need to perform another iteration.

4.2.7 Iteration 7

Having examined all applications from the database in the previous iterations, the approach in this iteration transitions to conceptual-to-empirical, while maintaining consistency in all dimensions. Since nothing changed in this iteration and all dimensions and characteristics are unique, all objective ending conditions are met. Further, the final taxonomy achieves a balance between conciseness and robustness as well as being comprehensive. Together with the explanations given in this article and the possibility to add new dimensions and characteristics, the taxonomy also meets the subjective ending conditions. Hence, the development process is concluded.

4.3 Final taxonomy

The final taxonomy consists of 15 dimensions, as shown in Fig. 4. Black cells indicate that the app is categorized under a specific characteristic, while white cells denote the absence of such characteristic. Our classification includes the 18 applications that were screened in detail for the taxonomy creation as well as the total number of appearances of specific characteristics amongst all 141 sports apps from two app stores and all 59 research apps.

Fig. 4
figure 4

The final taxonomy depicting the characteristics of 18 representative commercial VR applications for the development of the taxonomy

Full size image

4.3.1 Hardware

During our app store screening we observed that most apps run on the HTC Vive, followed by the Meta Quest, while only a minority is available for other headsets. While most of the apps used controllers for input, only a few enabled hand tracking. External devices, such as sports or fitness equipment, were rather rare. Most of the apps tracked the whole room, called roomscale, while some applications tracked the users’ hands.

4.3.2 Application purpose

The identified applications cover a wide range of sports. Most popular are ball sports that do not include the usage of bats, rhythm games, racket sports, bat-and-ball sports, and exergames. Overall, most of the remaining popular sports are also covered by VR applications. While commercial applications did not specify a target group, research applications targeted different age, fitness, and health levels. For those applications explicitly mentioning a gaming position, standing is encountered most often while a minority of applications allow sitting. The gameplay variation that is most popular is level diversity, meaning that the user can choose between different levels, e.g., different songs in rhythm games. However, in some applications the users can follow a story or career, which we call narrative progression. Only a few applications offer difficulty scaling within their levels.

4.3.3 Application design

Most of the applications implemented training facilities or nature environments as gameplay sceneries. The third most common scenery we encountered were sport venues. Aside from those natural sports environments, other applications implemented futuristic, fantasy, or sci-fi environments. Urban and empty sceneries were rather rare. While most of the commercial applications include gamification, research applications do not.

4.3.4 Game entities

Applications implementing avatars typically relied on athletes. However, some applications also implemented torso-only, minimalistic, animal, or casually dressed avatars. In the category of other avatars, opponent(s), spectators, and teammates are commonly encountered, while a minority of applications implemented instructors.

4.3.5 Payment

Most of the commercial applications need to be purchased once to play them, which we call pay-to-play. Only a small number of apps are either free-to-play or require a subscription. Some commercial applications additionally offer in-app purchases, either for content, cosmetics, or song packages.

4.3.6 Social interaction

A minority of applications allow multiple players to interact during the games. However, some of the commercial applications include direct competition, meaning that players are within the same environment and can see each other. Some commercial applications also implemented indirect competition, e.g., score boards where players can compare themselves to others despite competing on their own.

5 Discussion

In the following section, we critically examine our findings from the literature review and compare them with our final taxonomy. This comparison allows us to identify discrepancies and similarities between research and commercial VR sports applications, providing a differentiated understanding of the current approaches. Further, we evaluate whether current VR sports applications are designed in a way which facilitates learning and discuss technical challenges related to VR headsets. We then explore potential future research avenues and reflect on the limitations of our study.

5.1 Comparison of theory and practice

Our investigation into VR sports applications reveals that VR sports applications in the literature and in the app stores share several similar characteristics while also showing a range of differences which are depicted in Fig. 5. Our criteria for the comparison of theory and practice derive from the final taxonomy and the occurrence of different characteristics. Both literature and actual applications mainly run on two HMDs, the HTC Vive (31 research apps and 91 commercial apps) and Oculus/Meta devices (16 research apps and 73 commercial apps), like the newer Meta Quest which are prevalent due to their technological advancements and market penetration. This is congruent to headset statistics, e.g., by Steam. These devices are favored for their high-resolution displays, precise tracking, and large developer ecosystems, making them the devices of choice for delivering immersive VR experiences.

Fig. 5
figure 5

Venn diagram of similarities and differences between literature and commercial VR sports applications

Full size image

Further, the usage of handheld controllers is very popular in theory and practice, as a consequence of the maturity of this input method (29 research apps and 141 commercial apps). Controllers offer haptic feedback, which is essential for the tactile sensations that improve the VR experience, especially in sports applications where physical engagement is a key factor. Hand tracking (4 research apps and 2 commercial apps), while promising, has yet to achieve the same level of precision and reliability, although it offers possibilities for future innovation in VR interactivity. Note that some HMDs, such as the HTC Vive, are limited to controller input only and do not support hand tracking. Studies also showed that users feel a higher arousal and dominance when using controllers in comparison to hand tracking (Voigt-Antons et al. 2020).

Another similarity is the implementation of avatars for the users themselves (18 research apps and 50 commercial apps) and other entities, such as teammates, opponents, spectators, and trainers (46 research apps and 78 commercial apps). Avatars play a crucial role in both research and commercial applications enhancing the sense of presence in the virtual environment. In research applications, avatars are often designed to replicate the real-world movements of the users. This is particularly important for applications focused on skill acquisition or rehabilitation, where accurate movement replication is essential for immersion. Commercial applications, in addition, may often employ avatars for their motivational and engagement value, offering options to customize their avatars allowing users to personalize their virtual representation. This can lead to increased user attachment to the VR application and a more enjoyable experience which is crucial for keeping long-term engagement. Additionally, sharing virtual environments with other people is a key characteristic that classifies virtual reality (Walsh and Pawlowski 2002). As such, sports such as soccer, basketball, or table tennis rely on other players each with their own avatar. However, even for sports that do not require the usage of other players, additional characters can enhance the VR experience. For instance, the presence and feedback from an audience increases motivation and leads to better performance (Yu et al. 2023).

Aside from these similarities, we could also identify various differences. The first aspect only present in research applications is the definition of target groups (38 research apps and 0 commercial apps). Research applications target specific user groups in their studies, such as athletes, the elderly, or individuals with mobility impairments, and tailor their VR experience accordingly. This targeting allows for controlled experimentation and the collection of specific data to assess the effectiveness of VR experiences for these particular groups in academic contexts. Research applications are more focused on scientific precision than on mass market attractiveness. Commercial applications, by contrast, aim to attract a broader audience, offering VR experiences that are accessible and enjoyable for users of different fitness levels, seniority, and interests. This could be due to the playful character as well as various difficulty levels and a variety of activities that allows users from any age and fitness to use these apps. From a business perspective, designing apps for a broad audience could increase the number of downloads and hence the revenue generated of actual applications. When acting in a market environment with strong competition, excluding specific groups by defining a target group on the other hand could reduce the profit, especially since the majority of applications are not free-to-play.

Furthermore, research applications often applied external devices alongside controllers, such as ergometers, Xbox Kinect, or sports equipment, like rackets and bats (34 research apps and 5 commercial apps). The integration of external devices in research applications is often motivated by the desire to simulate real-world conditions or enhance the precision of data collection in their studies. Such devices help users to realistically reproduce the feeling of conventional sport providing valuable feedback on the user's performance and helping the translation of VR training into real-world improvements. However, commercial sports apps are limited in their adoption of additional devices, possibly due to the additional costs for potential users on top of the hardware expenses for the headset and the application. Commercial applications focus on a seamless user experience that minimizes the entry barriers. Requiring additional devices could deter potential users from downloading the app. Hence, the potential benefits may not be worth the implementation expense (Caserman et al. 2022). By removing external devices as a requirement, developers can simplify the setup process, lower the expenses for user, and potentially increase the marketability of their application. The trade-off is a less specialized VR experience, but one that is more accessible to the average consumer.

Another aspect of commercial VR sports applications is the kind of training. In the literature, usually specific skills are trained. They typically focus on the competitive aspects of sports, aligning with the research-driven nature of these implementations. The goal is to use the capabilities of VR to improve particular aspects of fitness or rehabilitation in a controlled and measurable way. On the other hand, fun is the main aspect of commercial applications. Their objective is to provide an engaging and enjoyable fitness experience that encourages regular use and promotes well-being. Commercial applications often apply the concept of gamification (14 research apps and 129 commercial apps), which might be due to the popularity of gaming in VR (Metamandrill 2023). Gamification in sports apps can fulfill users’ needs for competence, autonomy, and relatedness (Bitrián et al. 2020) and can directly impact behavioral intentions (Habachi et al. 2024) and users’ mood (Huang et al. 2017). By integrating gamified elements such as point scoring, achievements, and progress tracking, these applications motivate the users intrinsically and provide a sense of accomplishment.

The last literature feature that stands out compared to commercial applications is the concept of social experiences (2 research apps and 48 commercial apps), e.g., sharing the environments with other players which is one of the main characteristics of virtual reality (Walsh and Pawlowski 2002). Recent studies have shown that making sports apps social can be an effective way to integrate sports routines in users’ daily life (Tu et al. 2019). The concept of social experiences in VR fitness applications is particularly promoted in commercial implementations. These applications often feature multiplayer modes, leaderboards, and other social sharing capabilities. The addition of these social elements to VR applications not only reflects the communal aspects of team sports, but also serves the human desire for social interaction. In addition, such commercial incentives aim to retain users in the long term. These applications offer a high level of replay potential and intrinsic motivation through features such as multiplayer modes, leaderboards, and achievements, leading to customer loyalty and hence potentially higher revenues. However, for applications in literature that focus on the training of athletic skills, the additional implementation effort of making the VR apps social could be unnecessary for their specific research questions. Research applications may include other avatars, such as coaches or opponents, but these are usually controlled by AI or are pre-recorded, and they lack the social interaction that human players provide. While this can be sufficient for the purposes of a controlled study, it does not utilize the full potential of VR to create social, and interactive VR environments that goes beyond the individual experience. Regarding commercial applications, VR sports applications often include competition, either directly or indirectly. This is linked to the inclusion of social aspects since the contact with other players within the VR applications often leads to rivalries. Such competitive features are not only fun, but also provide players with ambitious goals and benchmarks for success. Interindividual rivalry can lead to higher sports performance and motivation (Kilduff 2014) and could hence encourage users to purchase and download VR sports apps. Research applications may include competitive elements, but these are usually secondary to the main research objectives.

In summary, the comparison of research-based and commercial VR sports applications shows that both areas pursue different objectives. While research applications aim to support testing the potentials and limits of using VR in sports, commercial applications drive innovations and need to address broad population groups while simultaneously focusing on user engagement through gamification elements. Both approaches benefit from each other by inducing new trends and delivering scientific validation.

5.2 Accordance of VR applications with learning theories

The taxonomy developed in our study provides a structured overview of the characteristics of VR sports applications, which can be analyzed based on the cognitive load theory, the situated learning theory, and the self-determination theory.

The findings from our taxonomy show that both research-based and commercially available VR sports applications often use immersive and interactive environments to enable physical exercises. According to the cognitive load theory, the immersive presentation inside VR can reduce extrinsic cognitive load and enhance germane cognitive load by allowing users to focus on the task at hand (Sweller et al. 2011). The taxonomy reveals that applications vary in complexity, from various difficulty settings in simple rhythm games to diverse gameplay modes in more complex team sports simulations, meaning that developers implicitly address the intrinsic cognitive load by designing tasks catering to different skill levels. Examples for applications with different complexity levels are “ib Cricket” (P2) with a variety of different game modes, such as coaching mode, campaign mode, CO-OP, or Tournaments; or “Golf + ” (P7) which offers different spaces to practice, play mini-games, and golf on virtual courses. This finding aligns with the theory on handling cognitive load to improve learning and skill development on an optimal level (Haryana et al. 2022).

Our taxonomy also highlights the diverse range of realistic scenarios and contexts that VR sports applications offer, from sport venues to natural environments. The situated learning theory says that such authentic contexts can enhance learning by providing experiences that mirror real-life settings (Lave and Wenger 1991). Almost all analyzed applications tried to implement realistic settings for their sports, such as “Carve Snowboard” (P9) which plays in a mountain landscape with ski slopes or “CleanSheet Soccer” (P4) whose environment is a conventional soccer stadium. The inclusion of multiplayer and competitive elements, like leaderboards, supports situated learning by facilitating social interaction within the learning environment. This is consistent with our taxonomy's findings where multiplayer is a key component of many commercial VR sports applications, improving the training of skills through cooperation and competition (Scavarelli et al. 2021). Apps, such as “WIN Reality Baseball” (P1) or “Gym Class– Basketball VR” (P3) already include social interaction and support situated learning. However, in contrast to that, most VR research applications are single-player only which could limit the learning possibilities based on the situated learning theory.

The taxonomy's insights support the principles of self-determination, particularly the basic needs theory. Research-based and commercially available VR sports applications offer autonomy through various gameplay options and customizable experiences in cosmetics or difficulty. For instance, the applications “Golf + ” (P7) and “Gym Class– Basketball VR” (P3) both offer different avatars to choose from as well as additional cosmetics or content that can be purchased in-game. By addressing the need for autonomy, VR sports applications can promote intrinsic motivation and encourage users to engage in physical activities (Ryan and Deci 2000). The taxonomy also shows that applications provide feedback and track progress by collecting points for various actions, which supports the need of competence. The applications “The Thrill of the Fight” (P6) and “First Person Tennis” (P10) for example already implement such progress tracking. Moreover, the social interaction facilitated by multiplayer game modes in commercial applications addresses the need for relatedness. For examples, such as P3 and P7, refer to the previous paragraph.

5.3 Technical limitations of VR hardware for fitness purposes

Developing VR sports applications presents unique technical challenges that can affect user experience. While the popular HMDs, such as HTC Vive and Meta Quest, are being favored for their performance and large developer community, they also have several limitations. Wired headsets compromise on the user's movement while wireless models can lead to interaction latency issues due to lower computational power. Both issues may affect the user’s immersion, leading to motion sickness or safety concerns. The taxonomy shows that applications implement sport disciplines that focus their movement to the upper body like table tennis or limit the user input on the upper-body region for sports that normally require full-body movement, such as soccer. On the other hand, sports, such as karate, aerobic and gymnastics, with strenuous full body movements are largely absent. Due to their complex movements, it might be challenging to replicate them authentically in VR. Furthermore, while controllers are favored for their haptic feedback and ergonomic design, they can present challenges when simulating certain sports movements. Sports that primarily utilize hand movements, such as climbing, basketball or sailing present a possible mismatch when simulated with handheld controllers, since these activities typically do not involve wielding objects like rackets or bats. On the other hand, sports that utilize equipment handling, such as table tennis, tennis, or baseball, are easier to simulate. The controller grip is analogous to the sports utensil to a certain degree, therefore enhancing the perception for the users.

5.4 Future research avenues

Since our taxonomy revealed specific characteristics with limited research applications, we could identify future research avenues to potential research fields.

5.4.1 Effectiveness of VR rhythm games

Our taxonomy showed that current research predominantly focusses on cycling (8 research apps), soccer (6 research apps), and exergames (8 research apps). However, rhythm games, while not appearing in our final set of research applications, have gained significant popularity in app stores (26 commercial apps). Despite this, there is a lack of comprehensive studies investigating the impact of rhythm games on users’ health. Exemplary research questions for the future could include: How do VR rhythm games influence users’ health compared to traditional exercise? Which psychological effects do rhythm games have? Randomized controlled trials are suitable to compare VR games with conventional methods.

5.4.2 Theoretical validation of VR water and winter sports

Although a range of VR water (4 commercial apps) and winter sports (9 commercial apps) applications are available on the market, the existing research base remains sparse, with only two research applications focusing on winter sports. We could only identify one study examining skiing and one examining ice hockey. Given the popularity of both sports, alongside with other winter and water sports, these areas present a great research opportunity. A high potential of training water sports, such as diving or surfing, in VR lies in the possibility to offer a safe and controlled environment. Skills could be trained before entering dangerous situations and hence reduce the risk of incidents. Future research could explore how VR applications influence water and winter sport experiences, especially in improving skills and fitness. Specific questions could include: How accurately can VR replicate conditions and environments of water and winter sport? How can water and winter sport skills trained in VR can be transferred to real situations? Can the usage of VR reduce the risk of incidents for water and winter sports? We propose to use comparative studies and experimental designs to assess the potential benefits of VR in those disciplines.

5.4.3 Effect of gamification

Several studies have shown the positive impact of gamification in sports apps (Huang et al. 2017; Bitrián et al. 2020; Habachi et al. 2024). Since the research on gamified VR sports apps is rather limited (14 research apps), especially compared to a high number of commercial applications including gamification elements (129 commercial apps), additional studies would increase the understanding of whether insights from conventional sports apps can be transferred to VR apps. Specific research questions might be: How does gamification in VR sports apps influence user adherence and performance? Are there differences in the user experience between gamified and non-gamified VR sports applications? Future studies should employ randomized controlled research designs to answer these questions.

5.4.4 Influence of social aspects on user experience

Additionally, the integration of social aspects, i.e., playing with or against other users, represents a significant area for future research. Although many research applications include other avatars (46 research apps), such as teammates, opponents, or spectators, they are in all cases controlled by a computer and do not represent actual users. While many commercial applications already include direct or indirect competition (48 commercial apps), only two research apps included the possibility to share virtual sports environments. The potential for real time human-to-human interaction within VR sports applications is yet to be explored. Togetherness is a key features of virtual realities (Walsh and Pawlowski 2002) and can be a motivating factor in conventional sports applications (Tu et al. 2019). Future studies could explore the following research questions: How do cooperative and competitive multiplayer modes affect the user’s motivation? Is direct or indirect competition more suitable to motivate users of VR fitness? What is the influence of competition on physical fitness? All three could be assessed in randomized controlled trials, as the questions above.

5.4.5 Technical challenges and immersion

Future research and development efforts should focus on mitigating technical limitations of VR headsets to enhance the user experience, and thus the marketability and user adoption of VR sports applications. Hand tracking, while not being popular in research or on the market (4 research apps and 2 commercial apps), shows potential in addressing the immersion of hand or full body movements, but currently lacks the precision and reliability in comparison to controllers. Possible research questions that could be addressed through empirical studies are: How well does hand tracking translate physical movements into VR compared to controllers in the fitness-context? Does hand tracking enhance the learning of sport skills compared to controllers?

5.5 Limitations

We want to acknowledge the limitations encountered in our study. The first limitation is related to the focus we set on peer-reviewed articles while excluding all other document types. While this focus ensures a high quality of the results, other literature, that might implement further interesting VR sports applications, is omitted. This approach may have excluded valuable insights from conference proceedings and grey literature, which could provide a more comprehensive picture of the current state of VR sports applications. Further, we only included studies that specified the use of immersive HMDs potentially overlooking applications developed for other output devices which might provide valuable insight as well. As VR continues to evolve, future research should consider a broader range of devices to fully understand the diversity of VR sports experiences. Our third limitation refers to the literature search string. We decided to keep the search string as general as possible to include a wide range of sports. While intentionally broad to capture a wide range of sports, this may have missed studies that used more specific approaches or focused on niche applications. Refining the search string through adding specific kinds of sport or expanding the search to include additional databases might have led to more results. To compensate that, we included a detailed forward and backward search to gather further articles in the considered sports. Finally, some potentially interesting articles were not available to us, even after sending full text requests. It highlights the challenges of accessing relevant literature, which may have impacted the comprehensiveness of our literature review. Ensuring a better access to publications could reduce such limitations of future studies.

6 Conclusions

This study has made several significant contributions to the field of VR sports. First, we created a taxonomy in an iterative process which is based on results of a systematic literature review of VR sports applications. Building upon this knowledge, we conducted a comprehensive comparison between research applications portrayed in the literature and the actual landscape of VR sports apps available in commercial markets. Our analysis revealed both similarities and discrepancies, shedding light on the evolution of VR technology within the sports domain. We observed that the differences between research and commercial VR sports applications highlight the distinct priorities and constraints of each domain. While literature applications mostly focus on specific skills of well-defined groups in order to conduct scientific research under controlled conditions, commercial apps operate in a market-driven environment where user engagement and the viability of monetization options are crucial. Therefore, they are designed for a wide range of people and designed in a fun, motivating, and competitive way. Both contribute valuable insights to VR sports, with research applications improving our understanding of the potential usage of VR for health and fitness and commercial applications driving user approval and user adoption. Derived from this comparison, we could identify several areas that require further research. Another contribution of this study is related to learning theories. We showed that many VR sports applications are designed in a way which supports learning. However, social aspects have been neglected in research so far.

In conclusion, this study contributes to the understanding of VR sports applications by synthesizing theoretical knowledge with real-world observations. By identifying aspects of convergence and disparity, we provide valuable insights for researchers, developers and practitioners aiming to leverage VR technology for sports-related experiences.