-->

 

Testing A Low Cost Mobile Robot Using Inclination And Switch Interfaces To Support Play In Children With Physical Disabilities

María F. Gómez M.1, Javier L. Castellanos C.1, Adriana Ríos 2, Antonio Cruz2, Daniel Quiroga 2, William Rodriguez2 and Kim Adams1

1University of Alberta (Canada), 2Universidad del Rosario (Colombia)

ABSTRACT

Play is a medium for children to explore and interact with the environment through the manipulation of toys. Through play children develop cognitive, social and linguistic skills. Play in children with physical disabilities may be compromised due to their difficulties in reaching and grasping. Assistive robots have been used as a means for children to interact with the environment and to manipulate objects when they are playing. However, the robots that are available on the market could be expensive, and the majority of them are controllable only with one interface, narrowing the population of children that can use them. The objective of this study was to test a low cost mobile robot that can be controlled using: the inclination of the head, the inclination of a smartphone placed on the head or controlled with the hands of the user, and three switches positioned near the head. Four typically developing children and one adult with cerebral palsy used the robot to perform a matching task. Results showed that all participants could complete the task using all the interfaces. Future work will be focused on implementing an algorithm to calibrate the interfaces according to the range of motion of the user’s heads. Additionally, an adaptive algorithm could be used in case users get tired and their movements become different from the calibration.

INTRODUCTION

Play is a vital and natural activity in which children can learn and explore the environment through object manipulation [1]. Through play children can develop cognitive, social, and linguistic skills. Play is a fundamental right that must be granted and promoted for every child [2]. Play promotes learning, discovery, mastery, adaptation, creativity, and self-expression in children [3].

Children with physical disabilities may have difficulties when manipulating objects and participating in play activities due to their limitations, such as difficulties in reaching and grasping. These limitations can result in developmental delays across different areas, including sensory, motor, cognition, interaction, communication, and social development [4]. Children with physical disabilities are less independent and tend to become spectators in play rather than active participants [5].  

The use of assistive technologies, such as robotic systems, can help children with physical impairments to have control over activities, the environment, and objects, thus enabling access to play and demonstration of their cognitive skills [6]. Robotic systems, accessed via alternative access methods, allow children to perform actions such as reaching for objects, turning them, stacking them, and others [6].  Several studies have demonstrated the potential that robots have for children with disabilities to interact with the environment, to promote exploration and learning, to develop social and interactive behavior, to improve independence, and to engage in play [6]. One example by Kronreif and colleagues [7] is a remote controlled robotic system, called PlayROB, to help children with disabilities play with Lego bricks. This robot is a three-degrees of freedom arm with a gripper, and was controllable by joystick devices [7]. Authors reported that children enjoyed the play activity, and that independent play had a positive effect on their self-esteem. In another example, children with physical disabilities enjoyed interacting with a switch-controlled truck-like mobile Lego robot, which increased their attention to tasks and their social and communication skills [8]. Finally, the IROMEC is a mobile robot platform that can be controlled using a remote control, switches, a touchscreen or by detection of movements [9]. Children with disabilities became the main protagonists of the play sessions and were equally active partners when playing with their peers and the IROMEC.

The aforementioned studies suggest that the use of robotic systems could promote independent play in children who have physical disabilities. However, according to a literature review performed by van den Heuvel and colleagues [10] the control interfaces and commercial availability of robots to support play in children with disabilities is lacking. Most of the technologies found in the review had only one interface to control the robot, indicating that only a small group of children can use them. Though the IROMEC can support more than one interface, it is still not commercially available [9].

Creating a robot that can be controlled with different types of interfaces may be a way of meeting the demands of children with different disabilities [10]. Additionally, a robot that is commercially available and affordable is needed, so parents/caregivers and therapists could use them at home or for therapy sessions. This study is part of a larger project that aims to commercialize a robot to support children with physical disabilities to play. The objective of the present study was to test a low cost mobile robot and compare two inclination control interfaces. Additionally, another objective was to compare the inclination interfaces with a switch interface, which is a common interface to control robots by people with physical disabilities.

METHODS

Participants

Four typically developing children who were seven years old, three females and one male, participated in the study. The age criteria was seven because children older than five years have the necessary skills to control mobile robots using switches [11]. All children had no known physical or cognitive impairments and had no previous experience using robots. Only one child had experience using a remote control car.  Additionally, a female participant who has quadriplegic Cerebral Palsy and was 49 years old at the moment of data collection tested the interfaces. She had poor motor control, which made the manipulation of objects difficult.

Materials

A set of five colored paper houses and five blocks of the same color were used for a matching task. The development and the materials of the low-cost robot are described in a previous study [12]. To reduce costs in this study, a Bluetooth module HC-06 was used instead of the XBee module, and a Baby Orangutan microcontroller was used to control the motors, replacing the Arduino Leonardo and the H bridge circuit.

The robot was controlled with three interfaces: two inclination interfaces and switches. One of the inclination interfaces was implemented as a head/neck inclination interface and the other as an app on a smartphone. The movements of the neck and head commanded the robot to move, so flexion of the neck (tilting the head forward) made the robot go forward, and left and right lateral flexion of the neck (bending the head left and right) made the robot turn left and right, respectively. Both inclination interfaces were implemented to control the robot with the angles measured by an accelerometer sensor, as described in the previous study [12]. The head inclination interface used an Arduino Leonardo, a HC-06 Bluetooth module, and a MPU 6050 accelerometer. The interface was programmed using the Arduino IDE. The smartphone was a Samsung Galaxy S7 edge that implemented the interface as a mobile application implemented in Android Studio 2.3. The switch interface consisted of three Ablenet Jelly Bean switches connected to a PC via a Don Johnston Switch Interface Pro 6.0. The switches made the robot go forward, left and right. The interface was programmed using Matlab R2016b connected via Bluetooth to the robot.

Procedures

The head inclination interface was attached to a hat that was placed on the head of the participants. The angle thresholds for the inclination interfaces were determined according to the head’s range of motion of one child and the adult. The same angle thresholds were used in both inclination interfaces for all the children. Initially, we tried to place the smartphone interface on the head of participants, but we had complications due to the size and weight of the phone, so we decided that the participants would control this interface with the hands. The assumption was that since both inclination interfaces worked the same way, testing the mobile application by placing the phone on the head would probably have had similar results as the head inclination interface. For the switch interface, the switches were placed on mounting arms near the left, right and back of the head so that children could control the robot using their head. For the adult, the left and right switches were placed near each hand and the forward switch was close to the left side of her head.

The figure presents the set up for the matching task performed in the study. There are five-colored paper houses (blue, red, green, yellow and orange) and the low cost robot set in a hexagon-form. Counter clockwise, they are arranged in the following order: blue, red, green, yellow, orange, and the robot. The distance between the blue and the red houses is 40 cm, as well as for the yellow and the orange houses. The distance between the blue and the orange house is 100cm, as well as for the red and yellow houses. The distance between the green house and the robot is 100 cm. Additionally, the low cost robot is positioned facing the houses.
Figure 1. Set-up for the matching task
Before doing the matching task, participants had the opportunity to get familiar with the interfaces. For the inclination control, researchers showed the participants how to do the movements to control the robot and then the participants moved the robot around. For the switch interface, the children used their hands to press on the switches to understand how the robot moved, before using the head to press the switches. The smartphone was tested only with children because the adult had difficulties controlling her upper limbs.

To perform the matching task, participants first used the switches positioned by their head, then the head inclination interface, and finally the smartphone interface. The matching task consisted of delivering colored blocks to the same colored paper houses. The houses were placed on the floor along with the robot, which was positioned at a starting position as depicted in Figure 1. A block was placed on the robot, one at a time, and participants were asked to deliver the blocks to the corresponding colored houses using the robot. Once the participant reached the respective house, the robot was taken to the starting position by the researcher.

The sessions were video recorded, and the videos were analyzed to obtain the time that each participant took to deliver a block. The times were calculated from the moment when the researcher gave the instruction to the participant to deliver the block until the time at which the participant stopped the robot by the house.

RESULTS

A graph is shown presenting the average time that each participant spent in delivering each block to the respective house using the three interfaces: Switches, head inclination and the smartphone. Average time in seconds is along the y-axis. Participants are along the x-axis. Child 1(CH1) spent in average 8 seconds using the switches, 11 seconds using the head inclination and 9 seconds using the smartphone. For child 2 (CH2) spent in average 10 seconds using the switches, 16 seconds using the head inclination and 8 seconds using the smartphone. Child 3 (CH3) spent in average 12 seconds using the switches, 16 seconds using the head inclination and 15 seconds using the smartphone. Child 4 (CH4) spent in average 15 seconds using the switches, 14 seconds using the head inclination and 10 seconds using the smartphone. The adult spent in average 11 seconds using the switches and 12 seconds using the head inclination.
Figure 2. Average time per participant during the delivery of each block to the respective house (Ch: child; A: adult).
The angle thresholds for the inclination interfaces were different for the children versus the adult with motor control problems. The range of motion of the adult was smaller with respect to the children’s. Also, when the adult intended to bend her neck to the left or right her head also tilted forward. To avoid the activation of the forward command when she performed lateral neck bending to the left or to the right, it was required to make the threshold for the forward command greater. For this reason, the threshold to move the robot forward was 30 degrees for the adult and 15 degrees for the children.

Figure 2 shows the average time that each participant took to deliver each block to the respective house. Despite a problem that the robot did not go exactly straight since the motor did not have rotation sensors, all participants completed the task and delivered all blocks to the houses correctly.                                                

DISCUSSION

Children could successfully complete the task using the head inclination interface. However, children had some issues turning in the correct direction, especially when the robot was oriented differently from their point of reference, e.g. facing them, so tilting left made the robot turn to the child’s right. Also, when they turned in the wrong direction, the children usually kept on turning until the robot was oriented the way they wanted. This increased the time to complete the task. In addition, children often tried to rotate the head instead of bending the neck to the left or right, to which the robot did not respond.

The adult with disabilities had a hard time using the head inclination interface, especially commanding the robot to go left or right. When she tried to go left or right, sometimes the robot moved forward. So, even though the angle threshold to go forward was set quite high, higher than for the children, the robot did not move 100% of the times to the direction she wanted it to move. Despite this limitation and the fact that the robot did not follow a straight line while going forward, she completed the task successfully.

Visual analysis of Figure 2 shows that most of the participants spent more time delivering the blocks using the head inclination interface compared to the other two interfaces. Only one child spent more time delivering the blocks using the switch interface, child Ch4. This could be because the child needed more time getting used to the switches than the other children, since for the first houses he spent more time than for the last ones. Another reason could be because the child took more time than other children to compensate for the robot not going straight (e.g. stopping the robot and turning it in the correct direction). Results for child Ch1 are very similar for the three interfaces, she got used to the interfaces very fast and seemed to not have any problems while using them.

Two out of the four children had lower times using the smartphone than the head inclination interface and the switch interface. Thus, the assumption about both inclination interfaces having similar results was not correct. This makes sense since the children used their hands to control it, and typically developing children are used to using their hands to manipulate objects. As the use of smartphones is growing everyday it makes sense to incorporate this kind of technology and its built-in features to create interfaces. The limitation of size and weight will potentially be addressed as manufactures continue to improve their smartphones, making them thinner, lighter, and of different sizes. Implementing the interfaces into the smartphones could reduce costs making the assistive technology more affordable.

Making the threshold angles for the adult with disabilities different from the children made the head inclination usable for her. For that reason, an algorithm will be implemented as future work to assess and measure the unique range of the motion of each user. This algorithm will be used as a calibration procedure. An adaptive algorithm could also be useful when the user gets tired and their head movements change over time, then the interface could adapt to new thresholds. Also, another possibility is to implement the head-inclination interface  with continuous values to control the robot instead of discrete actions for going forward, left and right (e.g. as a proportional joystick-like function). This will give more freedom to the user to move the robot in the environment, e.g. the robot could move diagonally or at different speeds. Additionally, other low-cost interfaces will be developed, such as eye gaze, electromyography or electroencephalography signals to control the mobile robot. Also, we will develop other interfaces for smartphones that could be used to control mobile robots. If more interfaces are available, people with different disabilities can use the one they feel most comfortable using. In addition, rotation sensors will be used to synchronize the motors of the wheels, to make the robot go straight when going forward.

CONCLUSION

This study tested two different inclination interfaces with four typically developing children and one adult with cerebral palsy, and compared these interfaces with a switch interface. Participants controlled a low-cost mobile robot to do a matching task having one of the inclination interfaces (head-inclination) positioned on the head and the three switches positioned near the head. The other inclination interface was programmed in a mobile application for Android smartphones and was tested by the children using their hands. Results showed that the matching task took longer to do with the head-inclination interface than with the switch interface, however all participants could complete the matching task without major inconveniences. The inclination and the switch interfaces can allow children with severe physical disabilities to control a robot using reliable anatomical sites such as the head. Future work will be focused on the creation of an algorithm to perform the calibration process based on the range of motion of each user.

REFERENCES

[1]       Besio, S., Dini, S., Ferrari, E., & Robins, B. Critical Factors Involved in Using Interactive Robots for Play Activities of Children with Disabilities, Assist. Technol. Res. Ser. Challenges Assist. Technol. 2007 pp. 505–509.

[2]       United Nations, “Convention on the Rights of the Child,” 1989.

[3]       Ferland, F. Le Modèle Ludique, 3rd ed. Montreal, Canada: Les Presses de l’Université de Montréal; 2003.

[4]       Klein, T., Gelderblom, G. J., de Witte, L., & Vanstipelen, S. Evaluation of short term effects of the IROMEC robotic toy for children with developmental disabilities.  In Proceedings of the International Conference on Rehabilitation Robotics (IEEE). 2011.

[5]       Blanche, E. Play in children with Cerebral Palsy: Doing With-Not Doing To. In Play in Occupational Therapy for Children, 2nd ed., D. Parham and L. Fazio, Eds. USA: Mosby Elsevier, 2008, pp. 375–393.

[6]       Adams, K., Alvarez, L., & Rios, A. Robotic Systems for Augmentative Manipulation to Promote Cognitive Development, Play, and Education: Principles and Practice. In Robotic Assistive Technologies. Principles and practice, P. Encarnação and A. Cook, Eds. Taylor & Francis Group, 2017, pp. 219–260.

[7]       Kronreif, G., Prazak, B., Kornfeld, M., Hochgatterer, A., & Fürst, M. Robot assistant ‘PlayROB’ - User trials and results. In Proceedings of the  Int. Work. Robot Hum. Interact. Commun (IEEE): pp. 113–117,2007.

[8]       Cook, A., Adams, K., Volden, J., Harbottle, N., & Harbottle, C. Using Lego robots to estimate cognitive ability in children who have severe physical disabilities. Disabil. Rehabil. Assist. Technol. 2011. 6(4): 338–346.

[9]       Marti, P., & Iacono, I. Learning through play with a robot companion.  Assist. technol. 2011. 29: 526–533.

[10]     Van den Heuvel, R. J. F., Lexis, M. A. S., Gelderblom, G. J., Jansens, R. M. L., & de Witte, L. P. Robots and ICT to support play in children with severe physical disabilities: a systematic review. Disabil. Rehabil. Assist. Technol. 2016. 11(2):103–116.

[11]     Poletz, L., Encarnação, P., Adams, K., & Cook, A. Robot skills and cognitive performance of preschool children. Technol. Disabil. 2010. 22 (3): 117–126.

 [12]    Rincon, A. R., Torres, D. A. Q., Castellanos, J., Gomez, M. F., Cruz, A.M., & Rodríguez, W.R. Testing a mobile robot toy for children with disabilities. In Proceedings of the World Congress on Medical Physics and Biomedical Eengineering, Toronto, Canada. In D.A. Jaffray (Ed.). June 7-12, 2015, pp. 1163–1166. Champ: Springer International Publishing.