Advanced AI and RobotIcS for autonomous task pErformance

保持可再生能源和农业等关键行业的竞争力和领导地位，取决于人工智能和机器人技术日益成为相关技术领域广泛和不可分割的一部分，特别是在面临劳动力短缺的情况下。ARISE项目旨在围绕可重构机器人机械手引入感知和控制模块的组合，这将使复杂操作任务的自动化水平发生飞跃。ARISE将包括以下关键的新技术组件，这些组件将在自动任务分割、人机交互和复杂操作方面显著推动最先进的技术水平：（1）安装在移动的机器人平台上的两个可重构的基于机械学的机器人操纵器（2）具有可变刚度的新型软末端执行器，其将允许不同的操纵任务（3）包括3D视觉算法的机器人感知模块（4）将允许机器人准确地识别其在环境中的位置的定位和映射模块（5）由场景理解算法驱动的语义映射模块（6）将捕获关于对象的属性和关系的重要信息的知识表示框架（7）一个分层模仿学习框架，用于获取机器人技能，直接从人类演示者那里完成复杂任务（8）一个人机交互条件路径和任务规划模块，实现反应式机器人控制（9）一个边缘AI框架，用于以流线型方式在边缘部署机器学习模型和计算机视觉算法。ARISE工具链将在5个真实的用例场景中进行集成和验证，包括分别在太阳能和水培农场中的安装和维修以及移植和收获任务。

Maintaining the competitiveness and leadership position of crucial industries such as renewable energy and agriculture, is contingent upon AI and Robotics increasingly becoming a widespread and integral part of the relevant technological landscapes, particularly in the face of steep labour shortages. ARISE project aims to introduce a combination of perception and control modules around a reconfigurable robotic manipulator that will enable a step change in the level of automation of complex manipulation tasks. ARISE will comprise the following key novel technology components that will significantly push the state of the art in terms of automatic task segmentation, human robot interaction and complex manipulation: (1) Two reconfigurable pneumatic-based robotic manipulators mounted on a mobile robotic platform (2) Novel soft end-effectors with variable stiffness that will allow for a diverse set of manipulation tasks (3) A Robotic perception module comprising 3D vision algorithms (4) A Localisation and Mapping module that will allow the robot to accurately identify its position within the environment (5) A Semantic Mapping module powered by scene understanding algorithms (6) A Knowledge Representation framework that will capture important information regarding objects’ properties and relationships (7) A Hierarchical Imitation Learning framework for acquiring robotic skills to accomplish complex tasks directly from human demonstrators (8) A Human-Interaction conditioned Path and Task planning module enabling reactive robot control (9) An edge-AI framework for deploying Machine Learning models and computer vision algorithms at the edge in a streamlined fashion. The ARISE toolchain will be integrated and validated in 5 real use case scenarios including installation and repair, and transplanting and harvesting tasks, in solar and hydroponic farms, respectively.