To clarify the subject, we will introduce definitions from the relevant literature to understand the term “urban service robot”. It consists of two generic terms to be further explicated. First, we clarify the type of robot we relate to, and second, the operational environment is specified.

A robot is understood as a “programmed actuated mechanism with a degree of autonomy to perform locomotion, manipulation, or positioning” (ISO, 2021b). The herein-mentioned term autonomy describes the “ability to perform intended tasks based on the current state and sensing, without human intervention” (ISO, 2021b). The norm differentiates between industrial, service, and medical robots. While industrial robots are defined as operating in an industrial environment, service robots perform tasks for humans or equipment. They are split into personal use, which includes non-commercial tasks and is usually done by laypersons, and professional use, which includes managing commercial tasks that require a trained operator (ISO, 2021b; International Federation of Robotics, 2022). Professional tasks include “inspection, surveillance, handling of items, person transportation, providing guidance or information, cooking and food handling, and cleaning” (ISO, 2021b).

Cities or urban places are described as having “a reasonably large and permanent concentration of people within a limited territory as the common characteristic” (Kuper and Kuper, 2003). Tiddi et al. (2020) emphasize three factors that are essential for such urban spaces, namely, the population, their relationships, and the existing facilities. Facilities can be understood as places frequented by the population for economic or social/cultural matters (Mumford, 1937).

Based on these definitions, we constitute urban service robots (USR) as professional service robots that are deployed in urban spaces and depending on their level of autonomy are operated by trained personnel. The higher the autonomy, the more the USR becomes a co-worker rather than a tool that is used or operated by someone. This distinguishes a USR from a common product during the use phase.

The term “sustainability” emerged from the domain of silviculture and was first brought up in 1713 by a German forester who described the importance of regulating the felling of trees, so it would not exceed the rate of the forest’s regeneration capabilities in order to keep a constant yield without destroying its source (Carlowitz, 1713).

The term was coined on a more general level by the World Commission on Environment and Development (WCED) in their report “Our Common Future” from 1987. They defined sustainability as “development that meets the needs of the present without compromising the ability of future generations to meet their own needs” (Brundtland et al., 1987). In creating this commission in the first place, the UN General Assembly reflected on the growing concerns about environmental problems on a global scale and its possible consequences for economic and social development. Although this is the most commonly referenced definition, it has undergone various interpretations so there is no generally accepted understanding of it (Thomsen, 2013). The long-term goals, referred to as sustainability, and the ways to achieve them, referred to as sustainable development, give room to interpretations, which may differ greatly depending on the conception (e.g., theory of weak or strong sustainability) (Jacobs, 2004). However, most of them share the idea that it requires the reconciliation of an ecological, social, and economic dimension.

The design process is part of the product life cycle and describes the development of a concept from an idea or assignment into a product ready for serial production (2221:2019-11). Autonomous service robots are highly complex mechatronic systems (Haidegger et al., 2013), and managing this degree of complexity in the development of a product requires a structured procedure. Therefore, the revised 2221:2019-11 divides the design process into several activities that will be run through during the process. It is important to note that these activities are overlapping and will be reiterated during the design phases. Each activity yields a certain result, catering toward the next activity or prompting a revision of a step before it ends in final product documentation. As stated in the introduction, we will center this paper around the early stages of design, meaning the first iterations of this process, including prototyping. During these stages, we focus on the “clarification of problem or task” and the “assessment and selection of solution concepts”. In between these steps, the product’s functions and its structure have to be determined for the creation of concepts suitable for a solution to the initial problem. Methods and tools to successfully complete these design steps are well documented in Gericke et al. (2021) and will not be touched upon in further detail throughout this paper.

We make this distinction because we deem the objectives and requirements as the basis for a product’s sustainability. They will shape the elaboration of concepts and hence codetermine the final solution. Secondly, the outcome of the selection process is particularly dependent on the applied set of criteria and their prioritization, which have to be adapted to favor sustainable properties and reach the desired result. The application of methods for assessment is not only suited to select among concepts but also to give directions and support decision-making during the concept phase (Wartzack, 2021).

Including social sustainability in the first activity (clarification of problem or task) is complex, especially with the product being a service robot operating in public spaces. The list of stakeholders has to be extended by including the people who frequent those spaces. They neither buy the product nor work with it but, nonetheless, have to share their space and are hence involved and an inherent part of its system or environment.

Therefore, we look at the early concept design phase and introduce an activity that we call preliminary social assessment (PSA). This process focuses on social sustainability and only includes the other dimensions when closely connected to social aspects (e.g., social acceptance can relate to demands for eco-friendly products or reflect on possible market potential). The main outcome, however, will provide social requirements with a focus on the service robot’s use phase. To complete social requirements connected to the remaining life cycle phases, the S-LCA provides key factors in that regard. Exemplary findings from the PSA and aspects from the S-LCA will be collected and converted into a checklist to support designers in this activity (see Table 4).

As seen in Figure 3, we create a phase diagram for a more detailed view. For our context specifically, we want to highlight two additional gates, one at the start and one at the end. The one in front is named the social responsibility gate (SRG) and the one at the end is the sustainability gate (SG). The SRG is derived from the concept of corporate social responsibility and functions as a first check-up of the incoming request or task to be aligned with the company’s own values. The SG concludes the development phase and determines how to proceed based on the gathered results. Both will be described further in the following Section 3.2.1.

For the collection of requirements within the ecological dimension, we provide another checklist generated from the LCA and Eco-Design categories to reduce the impacts of resource depletion, GHG emissions, and other damages to our eco-system. Concerning the economic dimension, we already showed this to be predominant in all relevant literature and norms (Section 2.2.1 and Section 2.3.1) and thus will not add to this subject. The aforementioned lists will complement the “clarification of problem or task” and add social and ecological next to the existent economic and functional aspects.

All four dimensions are crucial in forming a sustainable product. Some requirements can be related or dependent, so optimizing one of them might worsen the performance of another. The different solution concepts created during the design process reflect the impossible task of fulfilling all requirements optimally and vary in their pros and cons. Deciding on a solution with a rather equal distribution among the dimensions or leaning toward one of them is conducted in the assessment, evaluation, and selection activity. Originally, this is described in the activity “assessment and selection of solution concepts”, which in our opinion does not reflect the amount and variety of tasks and results that it contains so we suggest splitting it in two (see Figure 3). The first part contains defining and weighting the assessment criteria, whereas the second part holds the evaluation and selection of a concept. In both parts, we again integrate a social and ecological point of view to keep up with the four dimensions throughout the procedure.

The process chart in Figure 3 illustrates our suggestion of the activities’ sequence during the early concept design phase. For us, this phase signifies an extensive clarification of requirements and an assessment through first concept drafts and prototypes with a focus on all dimensions of sustainability, as opposed to the regular objectives orbiting around time and cost. We begin this section by describing the gates that we placed around the development phase, then explain the PSA and clarify the changes made to the existing activities at the end.

In the city of Berlin, Germany, the municipal waste management (BSR) maintains the public litter bins in the streets and parks, resulting in around 6.2 million emptying processes per year. The fleet dedicated to that task consists mostly of gasoline-fueled cars (two out of 49 were electrical). The project MARBLE by TU Berlin set out to optimize energy consumption and avoid CO2 and local NOx emissions. Therefore, we investigated how far a specially developed service robot can support and improve the process of emptying the public litter bins. The goal was to build a prototype to verify the technical feasibility and include all dimensions of sustainability in the development. The Project set out to focus on the ecological dimension with the condition not to disrupt the labor force while estimating the cost of such an endeavor. As the focus changed during the course of the project, we emphasized the process around the social dimension with a detailed description in the following section. Since the PSA evolved and was not ready to use from the beginning, we illustrated the steps in chronological order, which might deviate from the formalized steps that we discussed in section 3.

For an overview, Figure 7 shows a timeline of the project and the steps taken to investigate the social impacts of MARBLE. The initial concept idea for MARBLE was to operate on the pedestrian path and empty bins autonomously. This job was perceived to be done reluctantly and we aimed to free capacity for other tasks within the BSR as they struggle with personnel. So, the first activity was to investigate the basic conditions, the mechanical procedure, and the people involved. Therefore, we accompanied the BSR workers on their tour interviewing them on conditions and occurrent complications to get an overview of the operation sites. The main issues arising were the accessibility due to blocking obstacles or limited space. We also found that depending on the time of day and location, the amount and frequency of pedestrians vary greatly, which could influence the maneuverability and workflow. Also, the robot has to move on the sidewalk, limiting its size and speed. Workers usually empty around 200 bins during one shift and store the garbage in their trucks which are equipped with a compacter. For a small-sized robot, this requires some form of depot in close proximity to dump the collected garbage or a bigger-sized truck (mothership) that empties the robots. Additionally, the robots run on electrical energy to avoid local emissions, making it necessary to charge or change the battery. When asked about introducing a robot to their workflow, they seemed untouched and skeptical about the technical feasibility.

In the process of gathering requirements, it became clear that tools for social sustainability were not easily available within product development processes. Tools like the S-LCA or studies on the acceptance of SARs did not satisfyingly cover our case so we started drafting a first guideline tailored to the introduction of autonomous systems. While creating the guideline, we realized the necessity for an extensive description of the use phase and the embedding of the product into the process. This led to a first workshop, where we presented our drafts to the BSR (members from different departments) to exchange ideas and openly discuss concerns, obstacles, and improvements.

Since the service robot operates in public, it will move alongside pedestrians, including children, people with visual impairment or mobility aids, cyclists, or pets which need to be considered. There will also be forms of interaction with the BSR workers or technicians regarding control, battery change, garbage removal, service, or maintenance. Taking these stakeholders into account, we especially had to consider the time and place of operations. For example, operation during the night will entail the problem of noise disturbance and require the BSR to introduce nightshifts, as long as the system is not completely autonomous. Another big concern was vandalism and how to deal with it. Reflecting on such issues from different perspectives helped further the guideline and the second step we later called SIRA.

The first SIRA was conducted with a group of developers from different disciplines (mechanical engineering, environmental engineering, philosophy, business administration, and Engineering). Here, the fear of changing the job to a more monotonous position and the creation of hostile environments with unclear responsibilities was predominant. It was also noted that the implementation needs to be carefully planned to not replace workers but support them without changing qualification levels too much. As a benefit, the robot could alleviate physically straining tasks and increase job accessibility for people with disabilities. As a result, we wanted to include the stakeholders more in the development of the robot and embed them into the process, as our vision changed from autonomous operation to a concept of cooperation. The occurring interactions with people in public made research the topic of design concerning behavior, movement, and appearance. Human-like motions, audiovisual signals, and cute or human-like looks can increase acceptance and decrease fear or vandalism.

In the next step, we wanted to validate and analyze our assessment and first conclusions. A literature review on this topic confirmed the experience of little engagement during the very first interviews and early workshops. With the limited time frame of the project and one goal being the verification of the feasibility, we now concentrated on the development of a functional prototype to use for the step we later called acceptance analysis and validation of needs. For the same matter, a product designer helped create a design study implementing cute and human-like features. In the scenario building, we included the results from the SIRA and conceptualized two possible operation modes: the mothership scenario and the service station scenario. In both scenarios, a fleet of USRs supports the BSR in emptying litter bins. The first is complemented by the mothership, which is an electric vehicle, operated by a human to empty the robots and change empty batteries. Service stations will be spread around the city in the second scenario, where the robots can empty their compartment and recharge. Following the scenario technique, we extrapolated trends and assumptions to complete a picture for our USR in the near future. It consists of a growing city population and an increase in tourism which could result in a growing quantity of waste. The structural changes in the city could lead to more space for pedestrians and less space for private transport. The creation of service stations would benefit from this, but motherships could have difficulties reaching the service robots when streets get closed completely. However, operating a robot on pedestrian paths along with service stations demands space so a conflict could arise between pedestrians and robot operators. The advancement of technology might have a positive influence on the general level of acceptance of autonomous robots in public and reduce the danger of vandalism. With autonomous driving introduced, the legal framework already exists and autonomous USRs will easily conform.

With the design study (Figure 8B) and the first functions to demonstrate, we interviewed workers from the BSR and different groups of pedestrians (with mobility aids and visual impairment, as well as without a home). The interviewed pedestrians and workers underlined possible arising conflicts in high frequented areas during rush hour and again raised concerns of vandalism as inevitable. The general acceptance in their view is influenced by the efficiency and the outer design, which needs to be clear about the purpose of the robot. They both think it a good idea to start with test phases in more confined areas with less traffic like city parks. They had no concerns about safety and trusted that a deployed robot would be well-licensed and verified. The pedestrians noted that increasing the service of street cleaning could actually benefit the wellbeing of the citizens since they criticize too many full and overflowing bins. Important for them was to consider minorities and adapt the design accordingly. Interestingly, the perception of emptying bins diverges between workers and pedestrians. The latter see it as degrading and inhumane while the interviewed workers appreciated their job. The workers also highlighted the importance of the human—robot interaction to be easy and not stressful. They do not trust the technical feasibility yet, as it seems to be slow and inefficient. As a consequence, they fear the additional responsibilities of taking care of the robot and thus less time to concentrate on their original job. However, stating that older or injured colleagues are limited in taking on physically straining tasks, they see the potential for benefits by introducing a service robot. For the last workshops and survey, we first finalized the functional prototype (see Figure 8A) to demonstrate the robot operating the bin. The workshops were held with BSR personnel (four to six per workshop) from different organizational units (street cleaners, machine operators, repair shop staff, and route planners). It was important for us to do separate workshops keeping the managing staff apart from the street workers in order to avoid hierarchical influence on the discussions and statements made. The workshop began with a short introduction and demonstration of the prototype. After that, a moderator led through several topics (functionality, challenges, safety, acceptance, and impacts) inciting and guiding the discussions with the help of prepared questions. The discussion and statements were documented afterward. In the next section, a brief summary of their views is described. In the same phase, a survey was distributed among pedestrians during an open demonstration of the prototype in the streets of Berlin and we received 1131 replies. This helped to get their view on the robot since we had a limited capacity of workshops and aimed for higher participation rates.

In general, the workers from the BSR were very open to the idea of USRs supporting them. They could imagine taking over new tasks like changing the battery, handling some hardware maintenance, or steering it when stuck. These had to be easy-to-learn tasks and linked to their original job; anything related to software issues was excluded. Since there are already places with litter bins restricted to cars, they see advantages in a small robot. These are mostly park areas or squares, which the BSR sees as a suited test scenario anyway. They do not trust the robot to be effective in crowded streets except for nightshifts. Another input from the workshop was to adapt the robot and use it for other tasks like winter service or mowing of weeds. Furthermore, the feedback confirmed the potential benefit of the robot for the aging workforce by taking over physical straining tasks, making some activities more accessible to them. They disproved the fear of creating monotonous work since most workers are used to it and even appreciate it, as long as it will not confine them to a desk. We also found that the workers were predominantly curious about new technology and they expressed a certain pride to be part of the modernization. However, technical complexity could overburden some and they are afraid to be responsible for an expensive robot. Also, introducing an autonomous robot generates the fear of being replaced. They suggested limiting the number of robots so not all workers are obliged to adapt to the new tasks and it will feel more like an extension of the fleet.

The survey consists of statements connected to social aspects that have five different levels of agreement to choose from (e.g., strongly agree, somewhat agree, neutral, somewhat disagree, and strongly disagree). The participants were mostly in the age group of 30 and 49 (58%), 27% were younger, and 15% were above 50 years old. Generally, the attitude toward service robots like the one from the project MARBLE was positive and accepting. Nearly all participants strongly or somewhat agree (42% and 50%) that a robot can support the BSR and would feel safe around it. Interestingly, 78% said that they would accept service robots in urban spaces as normal in around 1–5 years, but they think it will take at least 5–10 years for the necessary technological maturity (76%). Particularly, the aspect “safety and accident avoidance” was chosen as the feature needing more development (50%). The impact of service robots on the urban infrastructure was seen as positive (42%) and somewhat positive (47%). However, the majority strongly or somewhat agree (45% and 46%) that it would change the pedestrian’s behavior and movement, hence this should be further investigated and considered in the development. The need for social and environmental regulations on the introduction of robots to public spaces was also agreed upon by nearly all participants: 43% strongly agreed and 53% somewhat agreed. The conceptual design of the robot was perceived as visually appealing although from several options, the participants favored modern design and human features over a cute look. The statistical analyses of the survey resulted in finding strong correlations between acceptance and the topics of sustainability, job quality for the BSR, design, and safety.

The workshops and survey took place at the end of the project and are ready to serve as input for the next iteration of a service robot with municipal tasks. Within the project MARBLE, this is as far as we got along with formulating these experiences into the PSA, hoping it will serve as a practical example and useful method.

In our first life cycle assessment, we compared our case study with the current fleet, consisting of 37 vehicles running on gas and 12 on electricity. The global warming potential (GWP) caused by the operation of this fleet impacted the whole outcome to clearly favor the use of our robot. However, since the BSR has committed to completely electrifying its fleet, we compare our two MARBLE scenarios with an electric fleet from the BSR in order to make this study valid for future comparisons. We also use the electric energy mix for the year 2030 based on the assumption of Agora Energiewende et al. (2022). The calculations for the results are based on the inventory analysis, meaning the gathering of data on all components and a simulation of the energy consumption. By measuring the energy consumption of the operations in the physical prototype, we validated the simulation data.

The inventory analysis is based on the prototype, presenting a first estimation. Gathering this data is cumbersome since most bought components, especially electronic devices lack information about the material composition and we could not afford to disassemble all parts for an exact identification. The used data for those parts rely on examples of similar components with partly size adjustments, making this another source of uncertainty. As mentioned before, the LCA in such an early stage will always come with uncertainties. It is therefore important to disclose these issues, as we might base decisions on the results. In the same manner, it is becoming common to use a 20% rule. It says that the difference between two scenarios, concepts, or products has to be above 20% to be deemed significant (Matthews et al., 2014).

In the category of GWP, Figure 9 shows that there is no significant difference between the mothership scenario and the reference, while the service station scenario produces the most (+40%). This is due to the fact that charging the battery results in more downtime for the robots and thus more robots are required to do the work at the same time. They also drive more distance to reach the service stations. The biggest impact is caused by the energy consumption (30%) during operation and the production (43%) due to the use of aluminum, copper, and rare-earth elements. In the mothership scenario, fewer robots are calculated by the simulation, hence the impact of production is 24% and consumption is 20%. The motherships’ production and operation take up around 20% each. The relatively low contribution of the use phase to the overall GWP (around 60%) is due to the energy mix calculated for 2030 with a high portion of renewable energy.

Next to the GWP, the LCA also looks at aspects like water consumption, land use, or terrestrial ecotoxicity (ReCiPe midpoints impact categories). Here the main drivers are the production of batteries and electric motors as well as the materials used for constructing the robots or electric vehicles due to the extraction processes. The mining and refining of, e.g., aluminum, copper, nickel, zinc, lithium, or chromium causes pollution and damage to human health by toxic substances used, fresh water shortage, and the destruction of landscapes and ecosystems. These impacts can be summarized in the endpoint indicators. While the results for the service station scenario remain the worst of all three scenarios, the mothership scenario was just slightly worse than the reference in damage to ecosystems (7%) and to resource availability (4%) but 5% better in damage to human health. Therefore, we can state that there is no significant difference based on the current data.

For the life cycle costing of our prototype, we included an estimation of the costs for software development, purchased parts, production of the robots, maintenance, and repair during the use phase. To put this into a reference for a conclusion, we took the number of emptying processes and calculated the current costs from the data given by the BSR. This comparison will only reflect the monetary costs of the service and neither touch upon any social-economic aspects nor suggest the replacement of human workers.

The current process by the BSR results in around 8 million € per year with the labor costs taking up 81% of the share. For a fleet of robots, the costs add up to around 6.3 million € per year. However, this estimation should be treated carefully. Especially, the use phase is unprecedented and, therefore, uncertain in its results. The personnel needed to maintain robot operation or the loss caused by vandalism cannot be put into valid numbers yet. Still, the LCC provides us with information on the distribution and most impactful factors. For example, the prototype has a robot arm to handle the complex motions to open the litter bin. Adapting the bins to robot operation and replacing the arm with a much simpler device yields an improvement of 20%. Another cost driver is the battery. Improving the capacity results in fewer robots needed and also less batteries used up per robot, resulting in an improvement of 15%.

We learned that it is very helpful for the PSA to have physical prototypes with a functional display of the tasks to engage the workshop participants. Although speculative design also helps to engage the participants, we found that in earlier encounters with no functional prototype, skepticism toward the technical feasibility was high. We observed that the statements made by participants got more concrete as they were facing the functional prototype. We assume that this is connected to the innovative nature of a USR, as it lacks similar existing products to be put in relation to.

Also, the gathering of feedback and comments by stakeholders is only the first step. Taking the information into account and translating it into design choices is a complex task and susceptible to misunderstanding. Therefore, the participation has to go on and complement the process. The results and conclusions have to be communicated transparently in a way that participants will be able to see their efforts put into action, as their suggestions might be neglected. The genuine intent to use tools for engaging stakeholders in the development process requires thorough planning and evaluation, as there are limitations and dynamics to be considered (e.g., stakeholder representation, appropriate mode of participation, execution, and documentation) (Felt and Fochler, 2010).

For the LCA and LCC, a basic concept of the robot and an understanding of its embedding within the whole process was necessary. Incorporating these evaluation methods, especially the PSA, will take up a lot of time and financial resources, which is an obvious drawback. The benefit lies in the documentation of the results and that they can be reused in modified and redesigned versions of the product. As our service robot is based on a modular design, the assessment of each module can be transferred or adapted. Building up the product structure early may result in extra work due to changes in the concept but will also help in the overall process of structured development. It remains questionable whether companies can afford this additional time and cost in the development of their products. However, we argue that this is necessary to arrive at a concept that considers all dimensions of sustainability and actually improves the result.

The data we gathered for the inventory analysis for the LCA can serve as the basic input for a complete S-LCA, which we did not conduct due to time and resource constraints. We rather focused on the use phase as this seems unprecedented. However, we learned from the LCA that the most impactful materials are almost equally used in the reference and robot scenarios. So, we can assume that the difference in those scenarios concerning the impacts discovered in the S-LCA will not be significant.

Within the scope of our project, we did not come to decide on a final solution concept. Particularly, the assessment of the ecological dimension showed that introducing a service robot yields no significant improvement compared to electrifying the entire fleet. One of the challenges is that some requirements connected to the ecological impact often counteract functionality issues or even ecological aspects themselves. For example, the choice of material is mainly bound to specific parameters to ensure stability and longevity of the construction and withstand occurring forces and weather conditions. Steel and aluminum fulfill these requirements but contribute to the GWP and a lot of the midpoint indicators in the LCA. In the same way, electric drives and the battery are made of resources that have a heavy influence on the ecological dimension. One way of improvement will be the reduction of weight. This will result in less material used and less energy consumed during operation. The motor and batteries have to be managed carefully and optimized for longevity. For instance, using the battery’s state of charge (SoC) between 10–70% increases the possible number of recharge cycles drastically in comparison to using the full range of SoC from 0–100%. Also, a redesign might help to reduce the number of drives and sensors needed, as we successfully prototyped a new way to operate the bins, eliminating the robot arm. More of these small improvements could refine the concept to score much better in the LCA and actually present a more ecologically sustainable alternative.

These suggested improvements are directly related to the economic aspects we evaluated. Fewer robots, less energy consumed, fewer materials required, and fewer electrical drives or sensors used directly lowers the cost of operation and therefore improves profitability. Going by the extensive definition of economic sustainability we supplied in Section 2, the robot should also improve the quality of life and services. As interviewed pedestrians criticized a lack of service regarding waste collection in the public space and the discomfort caused, our concept has the potential to improve service and quality of life by adding to the workforce. Of course, another way to solve this problem resides in decreasing waste generation. Although our scenario analyses deemed this outcome improbable, it is a more desirable solution to be discussed in another scope.

In terms of the social dimension, our concept changed throughout the process. We learned that the general idea of a robot working in an urban environment seems quite accepted under a few conditions. As replacement is the biggest fear of the employees, we need to make sure to find the right application that is either not taken, not feasible by humans, or unpopular. On the other hand, pedestrians mostly fear the conflict of space distribution so for the development of any USR, it is necessary to consider the time and location of the operation as well as incorporate possible structural changes in the city. In any case, the development process should be transparent and inclusive regarding the stakeholders. After the workshops, the participants expressed their appreciation and that the acceptance of such an innovation increases with each participation in the process.

Another topic is the comparability of the results and the scaling of assessment criteria. This holds especially true for the ecological and social aspects. The LCA results in quantitative indicators which makes them measurable and easy to compare to other scenarios. The problem here lies in the actual understanding of these indicators, as they are hard to grasp and intangible. Also, the individual indicators are hard to compare to each other and therefore hard to balance as they are measured in different units. Nonetheless, this method supplies the designer with information that can be used to reduce the impact. Clearly, we could identify the most impactful components or processes and tend to minimize them. In that regard, the method succeeds in considering the environmental dimension. As every product will impact the environment in one way or another, the question is: How much is acceptable and when should we as designers decide to discard a concept completely? However, how to go about the results is not within the scope of this method and probably needs to be determined on a political level.

The lack of standardized methods to incorporate the social dimension made us focus on this aspect the most during our project. The suggested PSA that we applied yielded many results and gave us an insight into possible impacts such a service robot might have. We are sure that it is impossible to foresee every impact on and reaction by society, but the amount of input we received by this method that otherwise would have been neglected shows how important and valuable this method is. The challenge inherent to the social dimension is that a lot of indicators are qualitative values that can neither be measured, expressed in units, nor compared to other criteria. We mentioned this in Section 3 and suggested a tool, which is used for the unitless comparison of functional and financial criteria.

However, this procedure cannot solve the fundamental problem of trade-off and the questionable transformation to match one unified scale. This is because the assumption to express all criteria in commensurable values implies the idea that the loss in one criterion can be compensated by increasing another (van de Poel, 2020). This might work for economic aspects but not for moral values (equality, quality of work, etc.) or criteria describing the pollution of freshwater or global warming. With our approach, we cannot offer a solution but want to underline the importance of being aware of this problem/dilemma. The first step to improve the assessment regarding this problem will be to look closely at the evaluation and not only the scoring since its representation should be questioned. Scoring from zero to four as the VDI2225 proposes can be based on the comparison to another product or scenario (dependent on ambition) or one’s understanding of social and ecological sustainability. As we said before, this is influenced by the corporate strategy and values of the company and the designers. The decision-making needs to be based on an interdisciplinary team that discusses the pros and cons. Our tool shall empower the ecological or social expert to have data and information to base their standpoint on and argue for or against a concept. The requisite is that there is such a team composition and that people in charge account for each dimension equally.

In Section 4, we already mentioned the problem of validity in such an early stage of the concept. On the one hand, flaws in the application or calculation can occur, and on the other hand, the aspect of uncertainty is inherent to the early use of such assessment tools. By updating the LCA and LCC twice during the project, we eliminated most errors. The uncertainties could not be eliminated but became clearer through the updates and can be categorized into three topics: The material composition, the energy consumption simulation, and the early stage of the concept. The first one refers to the lack of data concerning the material composition of purchased parts. This led us to assume the composition by taking similar products we could find in the database, resulting in uncertainties. The second is based on a virtual prototype and a virtual process simulation that cannot perfectly reflect reality and hence introduces another source of uncertainty. The last one expresses the uncertainty of the whole concept not being finalized. Components or modules might change and the embedding of the robot is not yet at a final stage. With this in mind, we once more point to the 20% rule (see sect. 4.3), yielding no significant difference in the ecological dimension but a significant improvement concerning the cost of the operation.

The LCC, LCA, and S-LCA are tools made for any product. The checklists we presented also originate mostly from existing tools intended for any sort of product. Additionally, checklists help to complete the list of requirements and are by no means mandatory, making them suitable for any product. Only the PSA has a special focus on the use phase and was developed along our case study of a USR. We see many use cases for municipal services that can be offered by such products. We see this method fit for such applications. On a higher level of abstraction, MARBLE is an automated mechatronic product. It is able to replace or supplement human workers as other machines already do in factories or might in the future of some service sectors. The PSA could in that case result in a different target group but still be useful to capture the requirements of those stakeholders. For products that do not perturbate public space or working conditions in the same manner, we believe it to be a too costly effort.