HUMAN-ROBOT COLLABORATION TNO, THE NETHERLANDS. 6 th SAF RA Symposium Sustainable Safety 2030 June 14, 2018 Mr. Johan van Middelaar

Transcription

1 HUMAN-ROBOT COLLABORATION TNO, THE NETHERLANDS 6 th SAF RA Symposium Sustainable Safety 2030 June 14, 2018 Mr. Johan van Middelaar

2 CONTENTS TNO & Robotics Robots and workplace safety: Human-Robot Collaboration, Interaction levels Robots and Cobots (AGV) Vulnerabilities Risk Control Options Elaboration of study results: Capabilities of Humans and Robots Co-active system model Model for Human-Robot Collaboration The next step. Future research (Japan EU) collaboration options

3 TNO & ROBOTICS Perceptual and Cognitive Systems: Human Robot Interaction Human Factors (interface design) Unmanned shipping Industrial Safety: Occupational Health & Safety Cybersecurity (Collaborative) Robot Safety Sustainable Productivity and Employability & Work Health and Technology Physical ergonomics Sustainable employability Exoskeletons & Telerobotics: i-botics

4 PROJECTS TNO INDUSTRIAL SAFETY For Ministry of Social Affairs and Employment: emerging safety risks w.r.t. workplace safety: 2016 (1) : cybersecurity: for working equipment connected to the internet 2016 (2) : robots: workplace safety and Human-Robot Collaboration 2017 : cobots: an exploratory study of AGV s at the workplace 2018 : cobots: risk control of AGV s at te wok place (not published yet) Literature studies, interviews with experts and workshops to identify the risks and the control measures. For each topic: public final report with knowledge map, mainly for end users. Reports available by the Ministry to EU Commission dealing with the revision of the EU Machine Directive 2006/42/EG (adjustment essential H&S requirements EHSR - Annex 1; after elections for a new EP in 2019 they will propose an initiative proposal for revision)

5 WORKING EQUIPMENT AND IT Study (1) 2016: Cybersecurity at the workplace Result Vulnerabilities of (future) working equipment w.r.t. IT: IT-Connections of equipment with the internet Remotely controlled working equipment Software failure Vulnerabilities Working equipment - IT Communication Costs of cyber security Rapid developments Outdated risk management Lack cyber-security awareness Lack of knowledge

6 WORKING EQUIPMENT AND IT Study (1) 2016: Cybersecurity at the workplace Results Risk Control Measures for working equipment Risk control - per life cycle (stage): 1. Design 2. Production & Systems Integration 3. Operation 4. Maintenance 5. Innovation 6. Demolition

7 DEFINITIONS OF ROBOT & COBOT TNO (working) definition robot: A robot is a machine that can be programmed, has sensors, and a certain degree of mobility, as a result of which the robot is able to carry out a task autonomously. TNO (working) definition cobot: A cobot or co-robot (of collaborative robot) is a robot that has been designed with the aim of having physical (and social) interaction with people in a shared work environment. This is in contrast to other robots, which are designed to work independently with limited interaction with humans. *In physical interaction, people interact with people using actuators (see robot definition) to intervene in the same shared environment (co-location).

8 HOW TO DEAL WITH COBOT RISKS? When can physical fences safely be removed? Which factors are of importance? How do we control H&S risks then? What role does too many trust or lack of trust in cobots play? What about the protection of human dignity in work (privacy, autonomy, wellbeing) T20XX What risks are involved with Artificial Intelligence and Machine Learning? What can the human operator decide or not?

9 HUMAN - ROBOT INTERACTION LEVELS Bron: J. Schmidtler et al. (2015) Human Centred Assistance Applications fort the working environment of the future.

10 HUMAN ROBOT INTERACTION LEVELS (Michalos e.o., 2015) Coexistential Cooperative Collaborative

11 HUMAN-ROBOT COLLABORATION TNO study (2) 2016: Robots: Workplace safety and Human-Robot Collaboration General: Definition of risk: Machine Robot Result 1: : Risk = f (likelihood; impact) : Risk = f (threat; impact; vulnerability) Summary Robots at workplace Vulnerabilities Vulnerabilities Robots at work place Change of human tasks (loss of skills) Unforeseen situations Trust in machines Shared responsibility Regulatory gaps Non-compliance Cyber security

12 HUMAN-ROBOT COLLABORATION TNO study (2) 2016: Robots: Workplace safety and Human-Robot Collaboration Result 2: Robots at work place - Risk control measures Risk control per life cycle (stage): 1. Design 2. Production & Systems Integration 3. Operation 4. Maintenance 5. Innovation 6. Demolition Risk control measures Design phase Assess robot functions (risk assessment, ) Involve end-users (buy-in,...) Review ergonomics Assess functional specifications (tasks, req s,...) Define user interface req s (knowledge, skills,...) Apply best available technology (sensors, ) Apply certification (machine, parts, ) Apply software testing (virtual, scenario s, )

13 COBOTS: AUTOMATED GUIDED VEHICLES (AGV) TNO study 2017: Cobots: exploratory study of AGV s at the workplace Example [igo neo, Still] Result: A model for Human Robot Interaction Basic principles: Human capabilities Robot capabilities Co-active system model

14 ELABORATION OF STUDY RESULTS (1/4) ROBOT CAPABILITIES (JOHNSON E.A. 2014) Human capabilities: Physical and cognitive workload Situation awareness Trust, acceptance and satisfaction Social emotional capabilities Human-Robot Interaction Relevant factors: : relation: function mental resources task human resources : perception of elements, comprehension of meaning and projection : human behavior related to robot (inter)action : binding, trust, communication, appearance and emotion : introduction, implementation, knowledge and training

15 ELABORATION OF STUDY RESULTS (2/4) ROBOT CAPABILITIES (JOHNSON E.A. 2014) Robot capabilities: Relevant factors: Perception : sensing presence of objects, movements Recognition : humans, (multiple) objects Predict : recognize (and remember), impact of actions, interpretaton circumstances Human-Robot interaction : physical, cognitive and/or social; design (protocols) Decision autonomy : sensors, self efficiency w.r.t. complexity of environment Transparancy/feedback : status, (mental representation of) intentions

16 ELABORATION OF STUDY RESULTS (3/4) LITERATURE REVIEW: CO-ACTIVE SYSTEM MODEL (JOHNSON E.A. 2014) Interdependence : set of complementary relationships that two or more parties rely on to manage required (hard) or opportunistic (soft) dependencies in joint activity. Inspirered by Clark s (1996) participatory actions and Fong s (2001) system model

17 ELABORATION OF STUDY RESULTS (4/4) LITERATURE REVIEW: CO-ACTIVE SYSTEM MODEL (JOHNSON E.A. 2014) Observability means making pertinent aspects of one s status, as well as one s knowledge of the team, task, and environment observable to others. Predictability means one s actions should be predictable enough that others can reasonably rely on them when considering their own actions. Directability means one s ability to direct the behavior of others and complementarily be directed by others. Inspired by Sycara & Sukthankar (2006)

18 MODEL FOR HUMAN-ROBOT-INTERACTION (Working) Environment Human - capabilities learning, experience, (team) knowledge, mental workload Observe: situation awareness, perception (e.g. perceived intelligence) Predict: communication and evaluation Direct: (decision) autonomy, adaptability, decision making, communication, team coordination Emotion: trust, acceptance, satisfaction Task complexity Critical tasks Responsibility Multitasking Surroundings Level of autonomy Safe Human- Robot- Interaction Robot -capabilities machine learning, experience, information ( knowledge ) Recognize ( Observe ): situation awareness, perception (e.g. perceived intelligence, emotion) Predict: communication and evaluation Direct: (decision) autonomy, adaptability, decision making, communication, team coordination

19 THE NEXT STEP FUTURE RESEARCH OPTIONS 1. Enhancing Situation Awareness with Persistent Robot-Assisted Disaster Response 2. Human-Robot-Interaction : development of HRI-capability profile (see next slide) 3. Experiment risk control measures in actual case study 4. Development of algoritms (design rules) for establising risk control measures in cobots software

20 HUMAN EPARTNER ROBOT COLLABORATION ontology policies simulations Situation Awareness Privacy Environment Work Load Team Coherence Safety Robots Model-based Knowledge Base learning Data-driven interactive thing interactive thing

21 POTENTIAL COBOT HRI PROFILE perception task complexity social interaction self-management 1. Observe recognize perception task complexity 2. Predict collaboration anticipate 4. Direct Self-management social interaction 3. Influence

22 BACKGROUND REPORTS Emergent risk to workplace safety as a result of the use of robots in the work place (TNO-report 2016 R11488) Safe application of robots in the work place - Safety chart Emergent risks to workplace safety as a result of IT connections of and between work equipment (TNOreport 2016 R11143) Safe work equipment and cybersecurity - Safety chart

23 THANK YOU FOR YOUR ATTENTION For more inspiration or information: TIME.TNO.NL