Reducing Vulnerability Regarding Care-Giving Robots

In contrast to in manufacturing, robots are still rare in supporting and interacting with humans. They have to some extent started to appear close to us for support in lawnmowing outdoor and vacuum cleaning indoor. Thus, it must be expected that progress will continue with more robots being close to each of us in the future. However, this is not only a technical challenge but rather an interdisciplinary one including humanities, social sciences, ethics, and law and policy. 

Bildet kan inneholde: robot, maskin, produkt, teknologi, elektronisk apparat.

In 2020, the ROBIN group starts a new Research Council of Norway funded project – Vulnerability in the Robot Society (VIROS, 2019-2023) ­– to address the vulnerabilities with regards to robots. The focus of the two full time researchers to be hired by the ROBIN group will be on privacy, security and safety, particularly in healthcare contexts. The project is a cross-disciplinary collaboration between the Department of Private Law and Department of Informatics at the University of Oslo and some other departments and partners. Its aims at in parallel developing technology and proposals for regulatory measures to reduce vulnerabilities regarding robotics.

Picture of TIAGo in the ROBIN common area

ROBIN has, during spring 2021, purchased a relevant robot, e.g. TIAGo from PAL Robotics, after considering several robots, including, Lio from F&P Robotics.The robots and their simulating environments can be applied in the master projects below. Similar to humans, the main functions of a robot are sensing, reasoning/thinking and motion control, respectively, which we would like to address in three different master projects (each linked to one of the three objectives in VIROS):

Below follows a number of different project description:

  1. Robot sensing: Multi-modal Sensing for Emergency Monitoring of Older People for Privacy
  2. Robot ethics: Machine Ethics for Robot Security
  3. Robot control: Motion Preferences for Home Robot for Safety
  4. Robot control: Reinforcement learning (RL) for safe human-robot interaction
  5. Robot safety: Exploring the concept of safety in care robots
  6. Robot design: An explorative study on Universal Design principles and care robots

  7.  Human-robot interaction: The theoretical and practical challenges with informed consent in AI-based care robots
  8. Human-robot interaction: The role of language-games in Universal Design: An explorative study about Human-Robot Interaction in care robots

1. Multi-modal Sensing for Emergency Monitoring of Older People for Privacy (Robot Sensing) 

With home care of older people, privacy need to be compromised with the ability of a system to notify the caregiver when some abnormal and emergency situation has happened. At the same time, light conditions in a home can vary a lot with sunlight at the daytime and darkness at night-time. Rather the transmitting a lot of personal sensor data to the caregiver facility, the local robot should train a model representing the regular activities of the person it supports. E.g., if the person does some daily exercises, a sudden increase in heart rate would be expected while it is no so for a person not usually doing any physical activity. Quality in the sensing can be improved in this way, see e.g.:

Ryo Saegusa, Hirokazu Ito, Duc Minh Duong, “Human-Care Rounds Robot with Contactless Breathing Measurement", in proc. of 2019 International Conference on Robotics and Automation (ICRA), Montreal, Canada, May 20-24, 2019 DOI: 10.1109/ICRA.2019.8794037

Goal of the project: Demonstrate how a combination of different sensors in a care robot at home can both increase privacy and at the same time increase quality in the prediction of emergency situations.

The tasks of the project:

  • Get an overview of earlier work on human activity recognition, human state monitoring and human behaviour modelling.
  • Plan and set up a data collection experiment where the subjects imitate various abnormal behaviours. 
  • Compare various machine learning algorithms for different set of sensors to see which combination is most effective.
  • Write master thesis.

Qualifications: Machine learning (e.g. IN3050/IN4050 and IN5490), robot control, programming, 

 

2. Machine Ethics for Robot Security (Robot ethics) 

This project is concerned with the question of how autonomous systems can be equipped with ethical values. Ethical autonomous systems are needed because, inevitably, near future systems are moral agents; consider driverless cars, or medical diagnosis AIs, both of which will need to make choices with ethical consequences. The challenge is how to implement ethical reasoning in machine. This journal special issue gives some recommendations: 

A. F. Winfield, K. Michael, J. Pitt and V. Evers, "Machine Ethics: The Design and Governance of Ethical AI and Autonomous Systems [Scanning the Issue]," in Proceedings of the IEEE, vol. 107, no. 3, pp. 509-517, March 2019. https://ieeexplore.ieee.org/document/8662743

This master project is concerned with in what way a home robot companion can be equipped with ethical reasoning capabilities to avoid misuse and security breaches. This is motivated by a number of examples on the Internet where people try to mislead self-learning systems.

The tasks of the project:

  • Get an overview of earlier work on machine ethics and security measures related to service robotic. 
  • Implement some demonstrations of machine ethics for a home robot companion and undertake experiments testing their effectiveness.
  • Compare the results to earlier and others work to improve and revise own proposed methods. 
  • Write master thesis.

Qualifications: Machine learning (e.g. IN3050/IN4050 and IN5490), robot control, digital security courses, programming 

 

3. Motion Preferences for Home Robot for Safety (Robot control)

This master project is concerned with addressing different ways a mobile robot moves around and how one or two manipulation arms are interaction with humans and objects. The motion speed and motion path will be a compromise between efficiency and actual and human-felt safety. This will to a large extent impact how humans trust a robot.

The tasks of the project:

  • Get an overview of earlier work on mobile robots and robot arm manipulation with regards to human–robot interaction. 
  • Plan and set up human–robot interaction experiments where the subjects experience different variants of robot motion patterns. 
  • Compare the results to see to what extent human-robot interaction is dependent on persons preference (e.g for serving food and removing remains afterwards)
  • Write master thesis.

Qualifications: Machine learning (e.g. IN3050/IN4050 and IN5490), robot control, programming, IN5000 – Qualitative Research Methods 

 

4. Reinforcement learning (RL) for safe human-robot interaction (Robot control)

This suggested project is proposed on behalf of the  Vulnerability in Robot Society (VIROS) research project. We propose a project that will investigate reinforcement learning for enhancing safety in human and robot interaction, focusing on care robots.The robot used in this master thesis is TIAGo. The thesis involves both simulating tasks and implementation on the TIAGo robot. Thus, where the experiments are done in both simulated and real environment, i.e., with real users. The suggested tools are Gazebo and TIAGo simulator. One of the following research questions can be explored in the project. The project can be adapted to the background and interest of the student, with a main focus one of the following RQs:

  • How can a robot in a safe way handover to a human different types of objects that require a high degree of accuracy (e.g., handover a glass full of water without spilling over, handover sensitive objects that can break if lost on the floor, such as eggs)?
  •  How can a robot move its arm to reach an object in a safe way when a human is in the vicinity of the robot? For example, how to avoid collisions between the robot and the human or static/dynamic obstacles in the environment? 
  • [Theoretical RQ that can be complementary to the practical part of the project: a) What is the definition of safety in human-robot interaction, and how to address it in executing HRI tasks? b)What are the advantages and disadvantages of using reinforcement learning for solving HRI tasks compared to other methods, e.g., control theory.]

Goal of the project: To work on reinforcement learning on a robot hand-over of objects to the human task. Imagine the object is for instance a glass full of water, a plate with small objects on it, a cup of coffee or eggs that need to be handed over to a person sitting in a wheelchair. The handover shall be done in a safe way.

The tasks of the project:

  • Understanding safety at a theoretical level.
  • Defining the exact problem, i.e., what safety indices will be addressed? Or which components of the robot will be considered in the definition of the problem? Only motions of the robotic arm (fixed robot base) or motions of the robotic arm and the mobile platform (moving robot base)
  • Learn the concept of reinforcement learning, perform a review about different RL methods, and select a method suitable for solving the safe HRI problem.
  • Design and develop the algorithm
  •  Learn Gazebo. Implement the developed algorithm on the TIAGo simulator.
  •  Design some experiments of hand-over objects from the TIAGo robot to the human that are implemented in practice and tested out with real users.
  • Write the thesis.

Qualifications:

  • IN5490 course (Advanced Topics in Artificial Intelligence for Intelligent Systems)
  • JUS5680 (Robot Regulation Course) (optional - if you wish to get equipped with some legal oriented theoretical skills regarding the concept of safety).
  •  IN5000 – Qualitative Research Methods  - for study with users.

References:                     

M. El-Shamouty, X. Wu, S. Yang, M. Albus and M. F. Huber, (2020) "Towards Safe Human-Robot Collaboration Using Deep Reinforcement Learning," 2020 IEEE International Conference on Robotics and Automation (ICRA), 2020, pp. 4899-4905

Quan Liu, Zhihao Liu, Bo Xiong, Wenjun Xu, Yang Liu, (2021), “Deep reinforcement learning-based safe interaction for industrial human-robot collaboration using intrinsic reward function” in Advanced Engineering Informatics, Volume 49, doi:https://doi.org/10.1016/j.aei.2021.101360

 

Bildet kan inneholde: ledd, datamaskin, bord, dataskjerm, teknologi.

5. Exploring the concept of safety in care robots (Robot safety)

Project description: The proposed thesis aims to explore the concept of safety in care robots at a theoretical and practical level. The concept of safety was so far associated with the physical layer of the robot. However, with long-term use of robots, the concept of safety may get new valences, such as cognitive safety. The research questions explored in this master thesis may include, but are not limited to:

  • What does it mean to include safety features in a care robot? What do we mean when we talk about safety?
  • What are some of the trade-offs that come with safety features (e.g., efficiency)?
  • How safe or unsafe is perceived an anthropomorphized robot compared to non-anthropomorphized robots? What design features of the robot gives the sense of safety to the human user?
  • What forms of human-robot interaction may contribute to the perceived safeness of a (care) robot? How should the robot behave in order to be perceived as safe and be safe?

This proposed master thesis is on behalf of Vulnerability in Robot Society (VIROS) and Predictive and Intuitive Robot Companion (PIRC) research projects. The robot and/or its simulating environment that can be used in this master thesis is TIAGo.

Goal of the project: The goal of the project is to explore the concept of safety.

The tasks of the project:

  • Get an overview of the existing work on safety and safe robots
  • Choose one to two research questions that you wish to investigate
  • Design and set up data collection with users
  • Write the master thesis

Qualifications

IN5490 course (Advanced Topics in Artificial Intelligence for Intelligent Systems)

IN5000 (Qualitative Research Methods)

JUS5690 (Robot Regulation) (Optional) – if the student wishes to focus on the legal side of the concept of safey, at a theoretical level.

References:

[1]                  ISO, “ISO 13482:2014, Robots and robotic devices — Safety requirements for personal care robots,” ISO, 2020. https://www.iso.org/cms/render/live/en/sites/isoorg/contents/data/standard/05/38/53820.html (accessed May 31, 2021).

[2]                  A. Martinetti, P. K. Chemweno, K. Nizamis, and E. Fosch-Villaronga, “Redefining Safety in Light of Human-Robot Interaction: A Critical Review of Current Standards and Regulations,” Front. Chem. Eng., vol. 0, 2021, doi: 10.3389/fceng.2021.666237.

 

6. An explorative study on Universal Design principles and care robots (Robot design)

Project description: This project aims at exploring Universal Design principles [1] at a theoretical and/or practical level within (care) robots. The student(s) working with this master thesis have the possibility to, for instance, illustrate the Universal Design principles applied to robots through prototyping [2], [3], by conducting user-studies, and/or by working with one of the following robots: T-ABLE, TIAGo robot or NAO. The proposed master thesis is proposed on behalf of the Vulnerability in Robot Society (VIROS) research project.

Goal of the project: The goal of the project is to explore Universal Design principles applied to (care) robots.

The tasks of the project:

  • Get an overview of the existing work in inclusive robotics/robots
  • Explore multimodal ways of interaction
  • Plan and set up data collection with users and the robot chosen and/or do the prototyping
  • Write the master thesis

Qualifications:

  • IN5590 Rapid Protoyping of Robotic Systems
  • IN5490 course (Advanced Topics in Artificial Intelligence for Intelligent Systems)

OR:

  • IN5000 (Qualitative Research Methods)
  • IN5510 (Participatory Experimental Design)
  • Tangible Interaction

References:

[1]                  Center for Universal Design, North Carolina State University, “Center for Universal Design NCSU - About the Center - Ronald L. Mace,” 2008. https://projects.ncsu.edu/design/cud/about_us/usronmace.htm (accessed Apr. 20, 2018).

[2]                  J. Herstad, T. W. Schulz, and D. Saplacan, “T-able: An Investigation of Habituating Moving Tables at Home,” Univers. Des. 2021 Spec. Mainstream Solut., pp. 238–251, 2021, doi: 10.3233/SHTI210400.

[3]                  D. Saplacan, J. Herstad, and T. Schulz, “T-ABLE - The Robotic Wood Table: Exploring situated abilities with familiar things,” Int. J. Adv. Intell. Syst., vol. 13, no. 3 & 4, Dec. 2020, [Online]. Available: https://www.iariajournals.org/intelligent_systems/index.html

 

7. The theoretical and practical challenges with informed consent in AI-based care robots (Human-Robot Interaction)

Project description: This project aims to explore some of the theoretical and practical challenges with informed consent in AI-based care robots. The theoretical challenges refer to ethical, regulatory or legal challenges with the implementation of getting informed consent in AI-based care robots. The practical challenges refer to technical or design challenges posed by the implementation of an informed consent set-up in an AI-based care robot that learns in a dynamic way from its interactions with the users. This master thesis is proposed on behalf of Vulnerability in Robot Society (VIROS) research project. The robot and/or its simulating environment that can be used in this master thesis is TIAGo.

Goal of the project: The goal of this project is to explore some of the theoretical and practical challenges with informed consent in AI-based care robots

The tasks of the project:

  • Get an overview of the existing work on informed consent and its practical implementation in AI based technologies
  • Set up data collection with users and/or with robots and users.
  • Develop and propose a theoretical framework dealing with informed consent in care robots.
  • Write the master thesis.

Qualifications:

IN5490 course (Advanced Topics in Artificial Intelligence for Intelligent Systems)

IN5000 (Qualitative Research Methods)

JUS5690 (Robot Regulation) (Optional – if the student(s) wishes to focus on the legal aspects of informed consent)

References:

[1]                  M. Coeckelbergh, “Artificial Intelligence: Some ethical issues and regulatory challenges,” Technol. Regul., vol. 2019, pp. 31–34, May 2019, doi: 10.26116/techreg.2019.003.

[2]                  European Commission, “Proposal for a Regulation on a European approach for Artificial Intelligence | Shaping Europe’s digital future,” Apr. 21, 2021. https://digital-strategy.ec.europa.eu/en/library/proposal-regulation-european-approach-artificial-intelligence (accessed Apr. 23, 2021).

[3]                  European Commission, “Proposal for a Regulation of the European Parliament and of the Council on machinery products,” Apr. 21, 2021. https://ec.europa.eu/docsroom/documents/45508 (accessed Apr. 23, 2021).

[4]                  UNESCO Digital Library, “Draft text of the Recommendation on the Ethics of Artificial Intelligence,” 2021. https://unesdoc.unesco.org/ark:/48223/pf0000377897 (accessed Aug. 10, 2021).

 

8. The role of language-games in Universal Design: An explorative study about Human-Robot Interaction in care robots (Human-Robot Interaction)

Project description: This project aims to explore multimodal types of interaction with care robots. Language-games is a philosophical concept coined initially by Wittgenstein [1]. If, initially, the concept addressed human to human communication, lately, the concept is explored in technology applications, in design [2], [3], or under the umbrella term of technology games [4], [5]. In robotics, language-games concept was used for robot-to-robot interaction, i.e., specifically for developing the robots vocabulary, or in robot swarms [6], [7].  At the same time, Universal Design principles have been explored so far in the physical/build environment, web- and mobile interfaces, however less in robot design and HRI, for a diversity of users. Roboticists are aiming at designing robots that can interact in multiple ways with multiple types of users, e.g., both children and elderly, persons with technical skills and with those with no technical skills. Furthermore, for the acceptance of robots in everyday life proper (non-) verbal communication of a robot is to be required [8]. To improve the naturalness of the communication between humans and robots, and thus the acceptance, rhythm could be a contributing factor [9]. This project aims to explore one of the robot language-games used in interaction with human users, either verbal (speech) or non-verbal language-games (motion, mimics or facial expressions, or gestures). How are the language-games perceived by the human in his or her interaction with the robot? Is the language-game perceived differently depending with whom the robot interacts with? Does rhythm play a role in non-verbal language-games?

Goal of the project: Explore verbal or non-verbal language-games at a theoretical and practical level in HRI.

The tasks of the project:

  • Explore current literature on language-games
  • Choose a verbal or non-verbal language-game to focus on
  • Set up a data collection study with users where the language-game chosen is explored together with TIAGo or another robot (e.g., NAO)
  • Write the thesis

Qualifications:

  • IN5490 course (Advanced Topics in Artificial Intelligence for Intelligent Systems)
  • IN5000 (Qualitative Research Methods)

References

[1]                  L. Wittgenstein, Philosophical investigations. Oxford: Basil Blackwell, 1968.

[2]                  A. Crabtree, “Talking Work: Language-games, Organisations and Computer Supported Cooperative Work,” Comput. Support. Coop. Work, vol. 9, no. 2, pp. 215–237, May 2000, doi: 10.1023/A:1008708914285.

[3]                  P. Ehn, “Playing the language-games of design and use-on skill and participation,” ACM SIGOIS Bull., vol. 9, no. 2–3, pp. 142–157, Apr. 1988, doi: 10.1145/966861.45426.

[4]                  M. Coeckelbergh, “Technology Games: Using Wittgenstein for Understanding and Evaluating Technology,” Sci. Eng. Ethics, vol. 24, no. 5, pp. 1503–1519, 2018, doi: 10.1007/s11948-017-9953-8.

[5]                  M. Coeckelbergh, “Technology Games/Gender Games. From Wittgenstein’s Toolbox and Language Games to Gendered Robots and Biased Artificial Intelligence,” in Feminist Philosophy of Technology, J. Loh and M. Coeckelbergh, Eds. Stuttgart: J.B. Metzler, 2019, pp. 27–38. doi: 10.1007/978-3-476-04967-4_2.

[6]                     L. Steels, “Language games for autonomous robots,” 2001, doi: 10.1109/5254.956077.

[7]                     N. Cambier, R. Miletitch, V. Frémont, M. Dorigo, E. Ferrante, and V. Trianni, “Language Evolution in Swarm Robotics: A Perspective,” Front. Robot. AI, vol. 7, 2020, doi: 10.3389/frobt.2020.00012.

[8]                     K. Dautenhahn, “Socially intelligent robots: Dimensions of human-robot interaction,” in Philosophical Transactions of the Royal Society B: Biological Sciences, 2007, doi: 10.1098/rstb.2006.2004.

[9]                     C. Breazeal, “Regulation and entrainment in human-robot interaction,” 2002, doi: https://doi.org/10.1177/0278364902021010096

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

                     

 

Publisert 6. okt. 2019 19:26 - Sist endret 22. aug. 2021 22:55

Omfang (studiepoeng)

60