Trustworthy Machine Learning by Demonstration

点击翻译按钮获取中文摘要

Bettering Our Worlds (BOW) Ltd. is working on developing software that will provide explainable, trustworthy and easy to use AI for commercial and domestic applications of robotic technologies. The goal of this feasibility study is to establish test-beds for this new technology in three different domains: domiciliary healthcare, dynamic manufacturing and hazardous environment operations.Within this project we will develop a hybrid machine learning approach that is capable of learning directly from skilled workers and provide explainability of decisions and actions. The system will not only feature training by skilled human operators but also integrates human-operated teleoperation as a failsafe for tasks that are too difficult or unpredictable for the AI.BOW believes that this approach will increase trust in automation, and overcome barriers to entry such as cost and deployment time. This project and its trust-centric approach is aligned with the long term vision and technological roadmap for BOW Ltd that has so far delivered a software development kit that enables anyone to code once and deploy to any robot. With the help of iUK we are bringing our expertise in democratising complex technologies and applying it to AI and ML to empower anyone to use these powerful technologies.