The AuRoRA Project studies the use of robots as tools that might serve an educational or therapeutic role for children with autism (Dautenhahn, 1999; Dautenhahn and Werry, 2004). The project was initiated by Prof. Kerstin Dautenhahn and started in 1998. Since then, a number of PhD students and postdoctoral researchers have contributed to this research.

Our main aim is to engage children with autism in a variety of ways helping them to develop and increase their communication and social interaction skills. Important social skills that we aim to facilitate include, but are not limited to, turn-taking, imitation, teaching about cause and effect, recognition of emotional expressions/gestures, teaching about appropriate tactile interaction, in addition to general communication and interaction skills that are required in human-human contact. Humans are the best models for human social behaviour, but their social behaviour is very subtle, elaborate, and widely unpredictable. Many children with autism are however interested in playing with mechanical toys or computers. Our project aims at children who have less developed communication and interaction skills.

Methodologically, our research has adopted case study, long-term evaluations with small groups of children with autism. Such long-term effects are important due to our commitment to address therapeutic and educational objectives.

Our research takes a strictly child-centred and assistive technology approach. We are not autism researchers, but try to help and assist children with autism in dealing with the social world. However, many of our studies involve collaborations with psychologists.

On the one hand we develop new robot technology and software for the robot to learn and adapt during the trials, on the other hand we investigate and develop new methods, methodologies and play scenarios for robot-assisted play and therapy for children with special needs, in particular for children with autism.

While the world may seem a complex and unpredictable place to a child with autism, the gradual integration into this world is an important factor in their ability to live a 'normal' life. The confusing nature of the environment, and social interaction in particular, often has the effect of causing the child to withdraw into a personal world. Autistic children seem most comfortable when playing repetitive games with simple toys and when they are alone. However, this environment may not be ideal for their development. The use of a robotic platform is an attempt to bridge the gulf between the stable, predictable and safe environment of a simple toy, and the complex and unpredictable world of human communication and interaction. People's social behaviour can be very subtle and unpredictable and may appear threatening to a child with autism. The use of a robot provides a simplified environment and can facilitate a gradual increase in the complexity of interaction, according to the individual child's needs and abilities.

In addition to investigating robots, we are also investigating how interactive software can access and develop story-telling and narrative skills in children with autism (Davis et al. 2007).
Our work comprises a number of research questions:

  • To develop different interactive games that children with autism can play with different humanoid or non-humanoid robots. These games are adapted to the child’s educational and therapeutic needs, and also depend on the particular behaviours that a specific robotic platforms affords.
  • To develop a robot control architectures for an interactive robotic platform, so that the robot functions as an interactive "toy" which based on a basic behaviour repertoire can express more complex behaviours, depending on the interaction with a child, or a small group of children.
  • To test the suitability of different robotic platforms, including mobile, as well as humanoid robots (Dautenhahn and Werry, 2000; Robins 2005). So far we have been using the mobile robots Labo-1 (donated by Applied AI Systems), Pekee (Wany Robotics) and the mobile robotic dog Aibo (Sony Corporation), as well as a small humanoid robotic doll (Robota, developed by Aude Billard), IROMEC- a mobile robot developed by a consortium during the FP6 European project IROMEC, and a child-sized humanoid robot KASPAR (developed by the Adaptive Systems Research Group at the University of Hertfordshire). We have been using fully autonomously operating robots as well as partially remote controlled robots (the control in the hands of the child or therapist/teacher/experiment who is part of the interaction scenario).
  • To develop learning algorithms so that the robot can identify and adapt to different play styles of children interacting with the robot (Francois et al. 2008).
  • To investigate different qualitative and quantitative evaluation techniques in order to point out the communicative competencies of children with autism. Quantitative techniques are based on the robot's sensors readings, as well as on an ethological analysis of video recordings of the children's play. Qualitative evaluations use Conversation Analysis (CA) for the in-depth study of behaviour in context. The latter approach has proved to be particularly useful for evaluating communicative behaviour that children with autism direct towards other people present during trials (carer, experimenter, or other children), (Robins et al, 2004a; Dickerson et al 2013). As part of this work we have also developed different experimental paradigms, e.g. the Theatrical Robot approach (Robins et al, 2004b).
  • To develop guidelines for researchers and practitioners for designing software and software trials for children with autism (Davis et al. 2009)).
  • To investigate how an autonomously operating robot can engage pairs of children with autism in games that foster collaborative play among children with autism (Wainer et al. 2008; Wainer et al. 2010, , Wainer et al. 2013).
  • To develop technology and play scenarios for children to explore socially appropriate tactile interaction and the relationship between tactile interaction and the expression of emotions. We published a set of 12 play scenarios, developed as part of the IROMEC consortium, suitable for children with special needs, including children with autism (Robins et al, 2012).
  • To highlight the added value of playing with a robot, compared to a non-robotic toy ( Werry and Dautenhahn 2007)
  • To develop a model of child-robot interaction where the robot acts as a mediator and an object of shared attention, encouraging the child to interact with peers (other children) and adults (teachers, carers, parents), e.g. Werry et al 2001(b), Robins et al 2005; Iacono et al, 2011)

The careful design and testing of communication and interaction techniques in robot-human interaction is a big scientific challenge for this project, and it can only be realised through a series of prototypes and their evaluation in interaction with children with autism. The project is therefore an ongoing long-term project.

The project collaborates with different schools in the Hertfordshire and Essex areas. Past studies were conducted at Radlett Lodge School, Colnbrook School, Bentfield Primary school, Middleton School and Woodland School (Edgware). Current on-going collaborations include Southfield school in Hatfield and Tracks (autism) in Stevenage.

In its very early stage, part of the project was supported by an EPSRC grant. Later, part of the research was funded by the European projects Robotcub, IROMEC and ROBOSKIN. Currently the project supported primarily by the KASPAR project which is funded by donations.