Robotic fitness teacher for children with Lego Mindstorms

This report summarizes a group project work for the course “Basics in Human-Robot Interaction ” by Martin Hammerschmid, Maximilian Lehr, Benjamin Stangl, Constantin Walcher  in the winter semester 2014/2015. It describes our motivation to build the robot, outlines the methodology for the proposed user study, gives an overview of the robots hardware and software, and concludes with the results from a formative evaluation with test users.

Motivation and related work

Children spend more and more of their spare time in front of digital devices such as smartphones, computers and televisions. Studies show that the current epidemic of obesity is caused by an environment which discourages physical activity [1]. This a major societal problem which affects the physical health and development of children. Physical inactivity is also linked to several serious health problems and medical conditions [2]. Encouraging physical activity is identified as an important aspect of preventing obesity in children [3]. While the reasons for this phenomenon is manifold, we also see that parents spend less time playing or doing sports with their children, which also poses challenges to the upbringing.

We believe that the physical embodiment of a robot could be a valuable media to especially engage children in playful physical exercises. Literature [4, 5] shows positive effects of exercise robots for the elderly, also indicating a strong user preference and longer engagement with a physical robot embodiment compared to a virtual robot [6].

As an inspiration of how to design the interaction between the robot and the children, we used the setup of typical workout classes in fitness centers: the instructor, in our case the robot, performs a sequence of movements facing the students, in our case the children. The children replicate the same movements live with the robot. The movements of the instructor and students are relative to the room (mirrored to each other), which is very intuitive when imitating movements.

Research questions

At the beginning of the project we asked ourselves the question: Which features of a fitness robot have a positive influence on the engagement in physical exercises that children perform together with the robot? We identified the following interesting features of the robot: speed of movement, direction of movement, sound, cooperative exercising, feedback stimuli.

For the scope of this project, we decided to focus on the speed of robotic movement and cooperative exercising. With our fitness robot we tried to answer the following research questions:

  • What effect do different speeds of movement by the robot have on the engagement of children?
  • What effect does cooperative compared to individual exercising have on the engagement of children with the exercise robot?

Experimental design

The robot acts as a “fitness teacher” demonstrating and instructing an interactive exercise program the children are supposed to mimic and complete (Figure 1). The exercise program consists of a set of movement tasks which are randomly recombined and altered. In the two-player mode the children exercise next to each other and compete who is faster at performing the indicated task (e.g. press the button of the robot or show a certain colour object to the robot).

Figure 1: Single child setup (left) and two children setup (right)
Figure 1: Single child setup (left) and two children setup (right)


The ideal participants in the study are healthy children (male and female) in preschool age between 4 and 6 years who are not underweight or obese. The children should be interested in participating in the study.

Measurements and test environment

The independent variables in the user study are the speed of robotic movement and the number of children exercising. The speed of movement will have the following two conditions: slower speed and faster speed. The conditions of cooperative exercising will be a single child and a two children setup. Combining the two variables resulted in a 2×2 experimental design with speed and number of children as dimensions:

  1. Slower speed of movement, single child.
  2. Faster speed of movement, single child.
  3. Slower speed of movement, two children.
  4. Faster speed of movement, two children.

The engagement (dependent variable) will be measured by the duration of each exercise session (time) and the total number of exercise sessions over the study period.

We suggest to perform the user study in a controlled but real environment such as a kindergarten or the gym of a primary school.

The documentation of the child-robot interactions will be by an unobtrusive video camera and an observer (Figure 2). The children will be filmed by a video camera on the side of the exercise area. An observer sitting behind the children will document and take notes of interaction errors or any other problems during the study. An instructor will give verbal instruction, a short demo of the robot, and start the robot via a remote control.

Figure 2: Test environment
Figure 2: Test environment

Robot prototype

The robot is able to perform an interactive exercise program at two different speeds within a defined exercise area. The robot is able to move back and forth to indicate directions of movement. The “arm” of the robot indicates movement of the users arms to the left or right. On the left side of the robot is a touch sensor and on the right a sensor detecting colour which are both part of the interactive exercise program.

The interactive exercise program consists of 15 randomly chosen subtasks. The robot can perform the following subtasks:

  • Follow the movements of the robot (forward / backward movement)
  • Move your arms to the left / right
  • Press the button of the robot
  • Show the yellow object to the robot and 
put it back to its original position
  • Show the blue object to the robot and 
put it back to its original position

Due to technical problems with the colour sensor, we had to exclude the tasks using this sensor from the exercise program with the test users.

Figure 3: Robot prototype
Figure 3: Robot prototype

Formative evaluation with users

The scope of our formative evaluation was to test the robot with three novel users. The goal was to see how the fitness robot performs and to draw conclusions about the exercise program, the robot hardware, software, and the overall study setup.

We identified the following problem statements for the evaluation:

  • Do users understand the purpose of the robot and engage with it?
  • Is the duration of the fitness program adequate (too short or long?
  • Is the fitness program interesting enough (too simple or complex)?
  • Do users react accordingly to the robot’s movements?
  • Do users find the sensors?
  • Is the hardware sturdy enough for several tests?
Figure 4: Examples of interactions with the robot in single-player mode.
Figure 4: Examples of interactions with the robot in single-player mode.
Figure 5: Examples of interactions with the robot in double-player mode.
Figure 5: Examples of interactions with the robot in double-player mode.

Evaluation results and recommendations

The points listed below summarize the our experiences from testing the robot with novel users and gives recommendations and improvements for the prototype. Our formative evaluation with three test users (age between 19 and 22 years) showed the following results:

  • The faster speed mode proved to be more interesting. Although we tested on a different age group than the target group, we expect that this would also hold true for younger users.
  • The two player mode was definitely more fun to play with. As a design recommendation, especially for robots with children, we suggest to always provide a two or multi-player mode.
  • The interaction modalities with the robot were intuitive. After a short demo, the users found the sensors and performed the program without mistakes or misinterpretations.
  • The duration of the program with 15 subtasks was not long enough. We suggest also for children to at least double the number to 30 or more subtasks. This can be easily achieved by setting a different number in the code.
  • The participants wanted to continue to play several rounds. However, in order for the exercise program to be engaging over an extended period, the robot would need additional subtasks. Therefore, we suggest to provide several levels with increasing complexity of the tasks and speed.
  • The hardware of the robot proved to be sturdy enough. The only issues we encountered was that the robot was too light. When the users pressed the button on the side of the robot, the robot turned to the side. We could fix this problem by putting the touch sensor on the front or the top of the robot.


[1] Hill, J. O., & Peters, J. C. (1998). Environmental contributions to the obesity epidemic. Science, 280(5368), 1371-1374.
[2] Consolvo, S., Everitt, K., Smith, I., & Landay, J. A. (2006). Design requirements for technologies that encourage physical activity. Paper presented at the Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, Montreal, Quebec, Canada.
[3] Goran, M. I., Reynolds, K. D., & Lindquist, C. H. (1999). Role of physical activity in the prevention of obesity in children. Int J Obes Relat Metab Disord, 23 Suppl 3, S18-33.
[4] Fasola, J., & Matarić, M. (2013). Socially Assistive Robot Exercise Coach: Motivating Older Adults to Engage in Physical Exercise. In J. P. Desai, G. Dudek, O. Khatib & V. Kumar (Eds.), Experimental Robotics (Vol. 88, pp. 463-479): Springer International Publishing.
[5] Matsusaka, Y., Fujii, H., Okano, T., & Hara, I. (2009). Health exercise demonstration robot TAIZO and effects of using voice command in robot-human collaborative demonstration. Paper presented at the Robot and Human Interactive Communication, 2009. RO-MAN 2009. The 18th IEEE International Symposium on.
[6] Kidd, C. D., & Breazeal, C. (2008). Robots at home: Understanding long-term human-robot interaction. Paper presented at the Intelligent Robots and Systems, 2008. IROS 2008. IEEE/RSJ International Conference on.

Leave a Reply