Robotic Safe Adaptation In unprecedented Situations

点击翻译按钮获取中文摘要

The robots of tomorrow will be endowed with the ability to adapt to drastic and unpredicted changes in their environment including humans. Such adaptations can however not be boundless: the robot must stay trustworthy, i.e. the adaptations should not be just a recovery into a degraded functionality. Instead, it must be a true adaptation, meaning that the robot will change its behavior while maintaining or even increasing its expected performance, and stays at least as safe and robust as before. RoboSAPIENS will focus on autonomous robotic software adaptations and will lay the foundations for ensuring that such software adaptations are carried out in an intrinsically safe, trustworthy and efficient manner, thereby reconciling open-ended self-adaptation with safety by design. RoboSAPIENS will also transform these foundations into 'first time right'-design tools and robotic platforms, and will validate and demonstrate them up to TRL4. To achieve this over-all goal, RoboSAPIENS will extend the state of the art in four main objectives. 1. It will enable robotic open-ended self-adaptation in response to unprecedented system structural and environmental changes. 2. It will advance safety engineering techniques to assure robotic safety not only before, during and after adaptation. 3. It will advance deep learning techniques to actively reduce uncertainty in robotic self-adaptation. 4. It will assure trustworthiness of systems that use both deep-learning and computational architecturesfor robotic self-adaptation. To realise these objectives, RoboSAPIENS will extend techniquessuch as MAPE K (Monitor, Analyze, Plan, Execute, Knowledge) and Deep Learning to set up generic adaptation procedures and also use an SSH dimension. RoboSAPIENS will demonstrate this trustworthy robotic self-adaptation on four industry-scale use cases centered around an industrial disassembly robot, a warehouse robotic swarm, a prolonged hull of an autonomous vessel, and human-robotic interaction.