The research project proposed here therefore consists of exploiting a decision-making architecture to program a mobile manipulator robot to perform machine cleaning and maintenance tasks in industrial environments, such as vacuuming and taking vibration measurements on machines. This robot must operate in shared autonomy, that is to say that the robot must be able to perform these tasks by itself at the request of an operator while having the possibility of being remotely operated. The proposed work will exploit the HBBA architecture to make the robot semi-autonomous while allowing it to be able to assess over time its level of achievement vis-à-vis its objectives. The robot must demonstrate decision-making robustness by examining how its behaviors are exploited over time in order to validate their proper functioning. If the robot does not appear to be performing its task remotely or autonomously, it must be able to use other remedies such as changing its behavior or notifying the operator.