Objective: the project "RobAutisTIC", works is conducted to evaluate the potential of a multimodal mobile robot to improve interactions and to study the child's social-communicative behaviours in this situation. The robot is used in a fun way to encourage interactive behaviours between the child and the therapist. We consider that child-therapist interactions are far more important than the child-robot interactions. The protocol we set up does not favour direct control of the robot by the child but only allows this control by the therapist. So, if the child wants to play with the robot, he/she must pass through a request to the therapist. The protocol uses 3 phases to i) demonstrate the robot's abilities, ii) show that the therapist may play with the robot, iii) encourage the triadic play. Main content: After a short introduction on autistic spectrum disorders, this paper quickly presents 5 projects involving robotics and autism followed by a brief section on the RobAutisTIC project. The main part focuses on the protocol, the behavioural coding, the experiments conducted with children and analyses a peculiar case and shows the facilitating action of the robot in the interpersonal interaction. Result: Most children are interested in the play and motivated. The main behaviour toward the robot is visual attention accompanied by other behaviours such as verbal/non verbal language sent to the robot. Conclusion: We cannot tell that this is the case for children who do not have a good level of verbalization; neither generalizes this analysis to all children involved in the experiments with the robot. Our future goal is to perform this analysis on several children with 3 experiments per child. This analysis will enable us, we hope, to demonstrate the possibilities offered by using a multimodal mobile robot in the interactions with an autistic child. In the longer term, we plan to quantify these interactions and to make the robot's behaviour adaptive.