Geert Pingen @gpgn

Cloud & Machine Learning Engineer, Research Scientist
Guide Robot


Handling transferring passengers efficiently poses a significant challenge in the commercial aerospace business. In 2013, the number of transfer passengers at Amsterdam Airport Schiphol - the largest airport in the Netherlands - rose to 22 million, totalling more than 40% of all passengers. It is clear that managing this massive migration of people effectively is of benefit for both the passengers as well as airliners. Passengers may get lost, have trouble finding their gate, or run into language barriers. This is just as small set of reasons for flights to be seriously delayed, and passengers missing their connecting flights.

A guide robot could be of help to passengers in need of assistance with finding their way around the airport, and it is exactly this purpose for which the SPENCER project was instigated. This EU funded project is doing research on the implementation of a robot platform that will be able to guide groups of transferring passengers at airport hubs, and by doing so decrease the current transfer time.


As part of this project it was the aim of this study to research the optimal behaviour of a guide robot, most notably in a crowded space such as an airport terminal. The interest of the study was to research the effect of head rotation and audiovisual feedback in guide robot behaviour.

In agreement with results from Butler & Agah (2001), Joosse et al. (2014) found in their research on Human-robot interaction performed with Chinese, American (US) and Argentine participants that, across cultures, users prefer a robot not to enter their personal space. Partially in accordance with previous studies on cultural differences in human interaction it was found that in the high-contact condition of Chinese participants, a closer approach was perceived as more comfortable. Argentinian participants however, though being a high-contact culture as well, responded more similarly to participants from the US, which favoured a less close approach. This could be due to Argentina being a relatively individualistic country compared to other South American countries. Participants found it more appropriate when the robot positioned itself between the mother and child in a triangular setup of mother-father-child with the robot approaching the group, than between the father and child. This could be attributed to the robot facing the father (alpha actor). This was taken into account in designing the robot's locomotor behaviour.

Shiomi et al. (2010) performed a guide robot experiment in the field in an attempt to attract more bystanders into overhearing its conversation, and thereby spontaneously participating in the tour. In the experiment, a guide robot was set up to greet, guide, and give advertising information to groups of people in a shopping mall. Results indicated that participants listened more when the robot was travelling backwards and that it attracted more bystanders to listen in, though the ratio of people being guided might be higher in the forward condition. This effect may be explained due to the area of audience being in front of the robot in the forward condition, meaning bystanders that are opposite to the robots movement direction have a hard time entering the area of audience. In contrast, when the robot is driving backwards, the area of audience is turned towards the group. This means that bystanders coming from behind the guiding robot can more easily enter the area of audience. Since airport guide robots should exhibit the same kind of behaviour as the tour robot in this study it was implemented, though care was taken to ensure that the rate of people arriving at the intended destination remains high enough, since attracting people should not be a top priority.

Robot platform

The mobile robot system used was a Festo Didactic Robotino v2.0. On top of the base system an MDF platform was constructed to attach the outer shell, putting the robot's total height at 170cm. This shell, which was made up of a metal frame and a combination of plastic and synthetic material, held a tablet which was used to display output to, and receive input from, the user.


The platform's hardware components included a stainless steel chassis where the power supply, the drive module (3 independent omnidirectional drive units mounted at a 120 degree angle to one another) and the distance and bumper sensors were located. On top of that a command bridge where the controller and I/O module were located;

Figure 1. Robot plaform interior diagram.

Figure 1. Robot plaform interior diagram.


The code running on the controller unit reads odometry data from the motors and Com (Command Bridge), and measures the distance travelled. The motors output the x/y coordinates with respect to origin. This data is sent through a socket to a TCP client running on the tablet, where the remaining distance is displayed after some post-calculations. Voice generation and recognition is done using the Google Android Speech library, relying on keyword recognition.


A between-subject study was performed with 25 participants to test the effects of robot head direction and rotation. The robot would guide groups of up to 3 people around an unfamiliar area at the University of Twente in different head rotation conditions accompanied by brief verbal dialog. Another iteration of the experiment (N=19) tested the optimal modality of audio-visual feedback during the guiding process.

Questionnaire results together with comment and video analysis revealed that participants evaluated the robot facing them and driving backwards as unnatural and intimidating. If information relating to the guide process is shown, backwards locomotion is rated more favourably. Furthermore, participants much preferred touch input to voice as a desirable way of interaction with the robot.

Figure 2. Participants engaging with the robot.

Figure 2. Participants engaging with the robot.

Further reading

Results were published at the International Symposium on New Frontiers in Human-Robot Interaction (2015).

A full publication is available at:


[1] Schiphol. (2014) Transport and traffic statistics. Schiphol.

[2] Spencer. (2011) European research project: Cognitive systems and robotics. Spencer.

[3] H. Hicheur, S. Vieilledent, and A. Berthoz, "Head motion in humans alternating between straight and curved walking path: combination of stabilizing and anticipatory orienting mechanisms", Neuroscience letters, vol. 383, no. 1, pp. 87-92, 2005.

[4] J. T. Butler and A. Agah, "Psychological effects of behavior patterns of a mobile personal robot", Autonomous Robots, vol. 10, no. 2, pp. 185-202, 2001.

[5] M. P. Joosse, R. W. Poppe, M. Lohse, and V. Evers, "Cultural differences in how an engagement-seeking robot should approach a group of people", in Proceedings of the 5th ACM international conference on Collaboration across boundaries: culture, distance & technology. ACM, 2014, pp. 121-130.

[6] M. Lohse, N. van Berkel, E. M. van Dijk, M. P. Joosse, D. E. Karreman, and V. Evers, "The innfuence of approach speed and functional noise on users' perception of a robot", in Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International Conference on. IEEE, 2013, pp. 1670-1675.

[7] M. Shiomi, T. Kanda, H. Ishiguro, and N. Hagita, "A larger audience, please!: encouraging people to listen to a guide robot", in Proceedings of the 5th ACM/IEEE international conference on Human-robot interaction. IEEE Press, 2010, pp. 31-38.

[8] K. M. Lee, W. Peng, S.-A. Jin, and C. Yan, "Can robots manifest personality?: An empirical test of personality recognition, social responses, and social presence in human-robot interaction", Journal of communication, vol. 56, no. 4, pp. 754-772, 2006.