There’s a new assistant in town, meet Polaris.

Usability Test

Since our field study at Aarhus University Hospital, we have been working on the interface for Polaris. We gathered a lot of insights from the field study which we have used in the design process to ensure that the solution solves a real need and provides a great user experience for the visitors of the hospital.

We discovered that it was difficult for visitors entering the hospital to make a decision on which of the many wayfinding aids they should use in order to get to their destination. Many of the aids does not lead the users all the way, but simply sends them in the right direction, meaning that users need to seek more information later on in their journey, maybe even from other types of wayfinding aids. With Polaris we try to make a personal guide that visitors can use all the way from the entrance to their destination. At the same time, we know from our previous studies that robots are great at capturing people’s attention, which will hopefully allow us to offer visitors the help they need, before they get confused by information overload from all the signs and screens. Once we have caught their interest, we try to make the interaction easier by letting visitors scan their health insurance card to identify themselves and find their destination, rather than having to remember or search for the destination in their invitation.

We also became aware of the important role of the volunteers, both as wayfinding guides, but also their empathic role of making the visitors feel welcome and reducing their stress or anxiety about their visit. We tried to translate some of these qualities into Polaris’ design, by mimicking a conversation and making sure that there is a high level of trust.

Together with the insights from the field study we also used our experience and knowledge from the airport use case in the design of the interface.

In order to solve potential problems with the interface before it is developed, we conducted a usability test where we got users to interact with the interface. The usability test, which was conducted at Combine’s office in Aalborg, was also a pilot study before going to Aarhus University Hospital to test a working prototype. We used a Double robot with an interactive prototype of the interface, on which the subjects performed some predefined tasks related to wayfinding. In order to efficiently get the most out of the test, we used a testing platform called Preely, which collects and sums up quantitative data, providing actionable insights about users’ behavior. Preely gave us insights on how the users interacted with the interface through data visualisation, including flows, actions, heatmaps, time per screen and user ratings. We enriched these results with qualitative data, by asking the participants to think aloud during the interaction, followed by an exit-interview. Augmenting our results with qualitative data helped us answer why users performed specific interactions, why something was misunderstood and how they experienced the interaction.

Test subject interacting with the Double robot

Test subject interacting with the Double robot

Test subject following robot to his destination

Test subject following the robot to his destination

After completing the usability test, an analysis of the gathered data was made in order to find potential problems and other important observations. 

All subjects managed to complete the tasks and find their way to the destinations with help from Polaris. The average usability rating of the interface was 8.25 out of 10. They said that this high score was primarily given due to the simplicity and ease of use of the interface. They also said that the visual design was very nice and calming, which might indicate that we have successfully implemented some of the volunteers’ qualities in the design. Furthermore the functionality they needed was present and easy to use.

However there was some confusion during one of the tasks for some of the subjects. This was primarily in regards to finishing the interaction after reaching the destination, or if they wanted to end the interaction before they reached their destination. The goal was to make Polaris seem polite and be proactive in offering further help before leaving the visitor, in case there was something else they needed. But the test showed that the subjects did not understand the purpose of these extra questions when they had already reached their destination. They found it more annoying than helpful, because they just wanted to end the interaction with Polaris quickly. In the hospital context, users often need to get to their appointments fast, and since the robot holds sensitive information about them, it is very important for them to make sure that the interaction is properly finished before they can abandon the robot. So what we thought would be perceived as a high level of service, actually turned out to be a nuisance. After the test we reconsidered the end of the flow and found a compromise where users can get the help they need, without being obstructed unnecessarily. The subjects also commented on some smaller issues with the interface, which we also tried to improve.

Beside the insights related to the interface, there were some interesting observations during the wayfinding with Polaris. It was observed that when the subjects followed Polaris to a destination some walked right behind Polaris, some walked obliquely behind and others alongside Polaris. These different ways of following the robot might reflect the perceived power distance between user and robot.

During the wayfinding, if Polaris came too close to other oncoming persons, some subjects felt responsible for Polaris’ actions and some even apologised on behalf of it. Furthermore some subjects said that when they reached their destination, they were in doubt if they should sent Polaris away or what they should do. It seemed that they felt a responsibility for Polaris, even after they had reached their goal and no longer needed the robot. This showed us that it needs to be more clear when the interaction with Polaris is done and what it will do afterwards.


Even though the test was not conducted in the real hospital environment, we still got some important and actionable insights. Because the interface has not been developed yet it has been fast and easy to implement the changes, which will hopefully make the experience with Polaris even better. After implementing these changes, the next step is to make a similar test at Aarhus University Hospital with the real Polaris robot and see how the visitors reacts to the meeting with the robot and how they interact with it.