In recent years, robots are being increasingly deployed outside strictly controlled environments. When faced with unexpected situations, these robots are often incapable of taking appropriate action and require human intervention. The goal of CONVINCE is to advance the capabilities of robots to perform complex tasks robustly and safely within unstructured environments via autonomous and unsupervised adaptation to the environment and operational context.
More specifically, the key contribution is to develop cognitive deliberation capabilities that ensure safe robot operation over extended periods of time without human intervention. These capabilities will be integrated into a model-driven software toolchain to allow developers to build application-specific deliberation systems able to i) determine robot’s behaviors required to fulfill a given task, also taking into account the context in which the robot operates and the experience gained during previous executions of the same task, ii) deploy and configure the components that are required to execute these behaviors, iii) automate the analysis of behaviors to ensure that they are safe and secure, leveraging on formal models and tools for design-time and run-time verification. The toolchain shall be based on proven system modeling concepts, particularly from the EU-funded project RobMoSys. Major parts shall be open-sourced with adapters to relevant robotics frameworks like the Robot Operating System (ROS).
To ensure real-world applicability, CONVINCE will demonstrate the technology developed in the project on three different real-world use cases, each of which presents unique technical difficulties and utilizes robotic systems of increasing complexity, in different application domains: vacuum cleaner robot, assembly robot, and robotic museum guide.