The ALMI project focuses on a novel proof-of-concept robotic assistant. The TIAGo robot is capable of speech interaction and object manipulation and can assist with everyday domestic challenges. 

The challenge

Service robots can perform many of the complex tasks required to provide assistance to people in their regular living environment. However, the adoption of robots in health and social care is limited due to safety concerns associated with their deployment in open-world environments. 

The research

This project will demonstrate how novel robotic technology, environment monitoring capabilities, verification techniques, and adaptation methods can be integrated and applied to address concerns for autonomous robots used in people’s homes.

A specific application is the focus of the research: a PAL Robotics TIAGo robot will provide long-term assistance through speech interaction and help with physical tasks to users with mild cognitive and motor impairments. The demonstrator is co-defined with a project partner who is a major dementia-care provider and will help advance robotics regulations and standards. 

The progress

The team has defined user privacy and regulation requirements for the demonstrator. Based on these requirements, the ALMI robotic assistive-care use cases for development in the PAL Robotics testbed have been defined. Additionally, research has started to develop a method for the modular stochastic modelling of the ALMI assistive-living scenarios using Markov decision processes, a modelling paradigm widely used in robot mission planning. This enables the modelling of both the aleatoric uncertainty inherent to the ALMI robot-human and robot-environment interactions and the non-determinism associated with the options available to the TIAGo robot at different stages of its assistive-care mission. These models will be used to ensure safe decision-making for the ALMI robotic assistive care demonstrator.

Recently, the team has built the first prototype of the new robotic arm for more advanced control and safe manipulation when collaborating with humans. This integration validates the work done since the beginning of the project within the robotic platform and allows for the demonstration of the system in a testbed in PAL Robotics.

The team have devised tool-supported methods for modelling the environment in which the robot will operate, at both low and high levels of path planning and navigation. 

Papers and publications

• Blog post about the ALMI project - Advancing robotics through EU collaborative projects
• Fang, X., Calinescu, R., Paterson, P., and Wilson, J. "PRESTO: Predicting system-level disruptions through parametric model checking". Paper available on Arxiv.

Project team

• PAL Robotics
• University of York