CORDIS - EU research results

Safety Enables Cooperation in Uncertain Robotic Environments

Periodic Reporting for period 2 - SECURE (Safety Enables Cooperation in Uncertain Robotic Environments)

Reporting period: 2017-09-01 to 2019-08-31

The majority of existing robots in the industry are pre-programmed robots working in safety zones, with visual or auditory warning signals and little concept of intelligent safety awareness necessary for dynamic and unpredictable domestic environments. In the future, novel cognitive robotic companions will be developed, which should be able to learn from users and adapt to open dynamic contexts. The development of such robot companions leads to new challenges for human-robot cooperation and safety, going well beyond the current state of the art. The SECURE project trained a new generation of twelve researchers on safe cognitive robot concepts for human work and living spaces on the most advanced humanoid robot platforms available in Europe. The main goal of SECURE was to prepare twelve young scientists for a career in academia and industry in the growing field of social robotics and human-robot collaboration. SECURE has successfully achieved this goal and, through collaboration with industrial partners and other MSCA actions, has given all fellows excellent opportunities for future employments and collaborations. SECURE fellows have already successfully transferred to jobs in top-tier international companies (Egor Lakomkin, Amazon Alexa; Chih-Hsuan Chen, Nobleo Technologies), have founded start-ups (Chandrakanth Bothe), are working on academic-industrial transfer projects with large companies (Mohammad Ali Zamani), or are continuing their research at top research institutes in Europe (Alessandra Rossi, University of Hertfordshire; Francois Foerster, INSERM Strasbourg; Grigorios Skaltsas, University of Hertfordshire; Mohammad Thabet, University of Manchester; Alexis Billier, Istituto Italiano di Tecnologia Genoa; Bruno Brito, Technical University Delft; Phuong D.H. Nguyen, University of Hamburg) and Canada (Marie Charbonneau, University of Waterloo). Therefore, the ideas and results developed within SECURE have reached all employment sectors, and innovative work has been carried out by twelve highly skilled researchers that are prepared to build the social robots of the future.
SECURE has organised in total seven events to deliver training top-down in the form of organised lectures and skill courses, and events organised by the peer-network itself with the help and guidance of the consortium. The largest events were a peer-organised PhD conference SSR-2018 in Madrid together with the fellows from the ITN SOCRATES bringing together young scientists in the field of social robotics, and the final workshop at the ICDL-EPIROB conference in Oslo in collaboration with the fellows of the EID A.P.R.I.L.
In terms of scientific results, the fellows have presented results beyond the state of the art in all three competence areas. The fellows in the “Embodiment” area have made significant contributions to improve the control and design of robot embodiments. The work encompassed both, the hardware design of robots, leading to more efficient and safe hands for collaborative human-robot tasks (Alexis Billier), as well as controllers, leading to a better 3D-awareness (Chih-Hsuan Chen) and full-body control of the robot (Marie Charbonneau, Bruno Brito) which is essential to improve safety when sharing environments.
In the “Situation” area, the range of research reached from neuroscience, where human behaviour when handling objects in certain situations was investigated (Francois Foerster), to developments where human-like peripersonal space was implemented to guide robot movements (Phuong D.H. Nguyen). In between, language and phonetic features were used to infer information about the current context in human-robot interaction (Egor Lakomkin, Mohammad Ali Zamani). Here, sentiment and emotional cues have been shown to enable a robot to improve its recognition of unsafe situations through the interpretation of human reactions.
The goal of the competence area “Interaction” was to extract cues from an interaction which can be used to judge the safety context of the interaction. The fellows have conducted successful research in this domain, investigating the extent to which user’s trust (Alessandra Rossi) and stress level (Grigorios Skaltsas) can be measured or influenced. Also, cues about the user’s emotional state can be recognised in a dialogue, including user’s sentiment and politeness (Chandrakant Bothe). And finally, neural models have been made more efficient to extract such data faster and more reliably (Mohammad Thabet). Altogether, robots can learn and monitor cues throughout an interaction to be able to detect sudden changes in the behaviour of the user who might hint at external events influencing the user’s well-being.
All results have been published in 40 peer-reviewed articles and have been presented to a wide scientific community at major conferences and workshops. Due to their visibility and expertise, SECURE fellows have already been internationally invited to give invited talks on their topics, have already co-hosted workshops, and have been involved in conference organisation. This together shows already the impact the fellows have on the scientific community. To reach the general public, invited presentations have also been given at different public events (Nights of Knowledge, Science Festivals). Robot demonstrations, where fellows especially focused on explaining the need for their research, have been given to all members of the public, from children (e.g. at the European Researcher Night) to adults and elderly citizens (at Open Days and lab visits).
With the increasing success of Artificial Intelligence, and especially the upcoming advent of social robots in health-related environments, research towards safe social robotics on all levels is essential. SECURE and its fellows thus have made significant progress towards this goal, by (1) providing insights into the interaction between robots and humans from a psychological and neuroscientific viewpoint, (2) improving the situation awareness of the artificial agents, including social features like sentiment and emotion, and (3) proposing novel methods for robot design and control, which in combination with the other two, have made a large step toward safe human-robot interaction in shared environments. Furthermore, all fellows have learned and understood the need to promote this idea of safe interactions as a key component of their future work and will continue to promote this idea in future research and outreach.
Through the interaction in SECURE and the secondments of fellows, also the interaction between the academic beneficiaries and the industrial partners has increased and has led to collaborative research outcomes and further project interactions, as already visible in the two joint events together with the other MSCA actions A.P.R.I.L and SOCRATES. Therefore, collaborations that have started within individual projects in SECURE have been developed to span several projects and partners, and will influence the collaborative research at the involved institutes for the coming years. The impact of the fellows, who are now spreading out into top positions in industry and academia, will be the spreading awareness among robot developers about the importance of safety in human-robot interaction, and the skills and experience on how to achieve it.