EN 
DE Systematically managing complexity
Authors: Prof. Dr. Martine Herpers, Fulda University of Applied Sciences, Department of Applied Computer Science, Robin Kirschner (BSc), Chemnitz University of Technology, Faculty of Mechanical Engineering
Contribution – Embedded Software Engineering Congress 2018
Research has been underway for some time on robots that can take over household tasks and offer support in caregiving. This paper presents a universal proposal for the systematic testing of the most important functional capabilities of eHealth service robots, based on practical experience gained in the residential laboratory at Fulda University of Applied Sciences and a modeling of the care robots and their living environment.
Introduction
Service robots have enjoyed increasing popularity in the home environment for years [1]. They take over tasks in the household and garden or assist in the care of elderly or sick people. The mobility of autonomous robots requires the ability to navigate through unfamiliar spaces using suitable sensors. The home environment places very high demands on mobile robots. What poses no problem for humans, such as closed doors, moved furniture, or diverse shapes, can be an insurmountable obstacle for robots [2]. eHealth applications also require that the robots be able to approach and communicate with people.
For a systematic test, the complexity of the robot and its environment must be adequately considered. Most tests are scenario-based and conducted in various households or facilities that are available more or less randomly. Since the home environment depends on the respective culture and the residents' affinity for technology, comprehensive testing is virtually impossible. Therefore, for example, testing a care robot in a Japanese household has only limited relevance for use in German living rooms.
methodology
With a detailed Literature review Furthermore, in our own studies, we were able to identify some crucial requirements for the test environment/living environment for autonomous mobile care robots [3] [4], which are presented below. The simplest version of mobile robots in the home are vacuuming or mopping robots, which must achieve a high degree of floor coverage. exploratory tests The robot's behavior in various situations, such as starting at the charging station, overcoming unexpected obstacles, and potential problems with furniture, was observed and systematized. For tasks performed on or with the eHealth client, the robot's accessibility to the client is crucial. Since 2014, various robots, such as Pepper, LIO, Car-O-bot, and ROBEAR, have been tested in care facilities, and their differing designs and features were taken into account in the modeling of a general model for a care robot.
Modeling care robots and home environments
The different robots can be generally modeled as follows. With regard to locomotion, they consist of a mobility unit (usually wheels), a body, and actuators/end effectors such as brushes or arms. The robot's dimensions, such as the maximum body diameter plus attached actuators (RK),max), the height (RH) and the range (RR) are crucial for usability in residential environments (see Figure 1, PDF).
following nine categories The following must be considered when defining a test strategy: narrow passages (P), hanging objects (hO), furniture with legs, unstable objects, hidden areas, reflections/light (including colors and transparent obstacles), drops, small loose objects on the floor (e.g., carpet fringes, power cables), and room shapes. The specific characteristics of these categories required for testing an eHealth robot depend on its shape, mobility unit, and the sensors and actuators used. The categories are briefly described below.
Narrow passages Narrow passages exist when the width b and length l of passage P correspond to formula (1). The variable factor ε must be chosen appropriately in each case. (See formula in PDF)
Narrow passages can become obstacles for robots that need to reorient themselves or perform movements within them. The longer the narrow passage, the more likely it is to become an obstacle for a robot.
Hanging objects Hanging objects are critical when they hang down to approximately the robot's height, i.e., when the distance to the ground (D) under the hanging object is slightly less than the robot's height. Stable hanging objects at this height can contribute to the robot becoming stuck. (See formula in...) PDF)
Furniture with legs Furniture with legs is always problematic when the robot can move under the furniture or when the mobility unit/actuator/end effector can get caught on the furniture legs. In the first case, a maze can form, which can limit the robot's effectiveness.
A obstacle is considered unstable (instable objects) should be considered if they can be unintentionally moved, damaged, or become a hazard to eHealth clients (e.g., a tripping hazard) through contact with the robot.
Hidden areas (Hiding areas) exist when eHealth clients cannot be reached by the robot at this location, i.e., when the minimum distance Dmin The robot's distance to the client is smaller than the robot's reach (RR). Such areas are often the favorite spots of eHealth clients, which are, for example, obstructed by side tables, or arise unintentionally, for example, due to dark carpets that the robot interprets as a precipice and therefore does not cross.
Reflections/Light This can interfere with robots that use optical sensors or cameras. Floor-to-ceiling windows and other reflective objects can cause unexpected lighting effects.
abysses These limitations depend on the robot's mobility unit. Most household robots and eHealth robots move on wheels that can easily overcome differences of a few centimeters. Larger differences in floor level or stairs represent insurmountable obstacles that must be reliably detected. Four-legged robots could overcome stairs, but are not currently manufactured for household use.
Small, loose items Robots often have difficulty detecting obstacles. For most household robots, the floor should be cleared before use. In particular, very small objects that can get under the mobility mechanism and between the wheels or legs can pose a danger to the robot.
Rooms can have very different shapes. For the robot, the space is further restricted by the furniture. Each Spatial form Corners can range from rounded to sharp.
The above-mentioned categories can be used for the systematic system test can be used to complement scenario-based testing. Depending on the dimensions and sensors/actuators of the robots, the Limit value analysis The height and arrangement of the furniture are determined, and it is established for each category whether it has been sufficiently tested. Classification trees The test cases can be clearly specified.
The Living Lab The living lab, located at Fulda University of Applied Sciences, Department of Nutritional Sciences, was used for testing simple eHealth robots. It offers an office, dining room, sitting area, hallway, and living room, separated by sturdy curtains. Floor-to-ceiling windows allow daylight to enter, which can lead to varying light conditions. Constant lighting can be achieved using the living lab's highly flexible lighting system. Most tests were conducted with the Roomba 680. The Turtlebot 2 uses an identical mobility unit and has the same diameter. The only difference between the two robots is their height. Their stable and reliable mobility units, along with their high acceptance in the home, make these robots ideal testing platforms. The living lab offers most of the categories mentioned above. (See Figure 2.) PDFThe narrow passages, hazards posed by unstable and hanging objects, and areas where the robots need to hide are marked. Reflections occur through the floor-to-ceiling window in the living area, and different lighting conditions can be created using flexibly positioned ceiling lights. Various floor coverings can be used.
Conclusion
Since the Living Lab does not include a kitchen or bathroom, further tests were conducted in private settings. Because low-hanging, fixed objects in bathrooms or toilets can cause obstructions to walls and radiators in the confined spaces, the Living Lab will be expanded to include such testing opportunities. The defined categories can help in the future to systematically test household and eHealth robots and serve as input for test-oriented robot development. Further projects will analyze the specific characteristics of different robot types in the care sector in more detail in order to recommend even more specific test scenarios and environments.
Link list robots
Pepper from Aldebaran and Softbank Mobile
Turtlebot by Open Source Robotics Foundation, Inc.
Bibliography
| [1] | J. Young, R. Hawkins, E. Sharlin, and T. Igarashi, „Toward Acceptable Domestic Robots: Lessons Learned from Social Psychology,“ 2008. [Online]. Available: hdl.handle.net/1880/46693. [Accessed September 11, 2018]. |
| [2] | HM. Gros, S. Mueller, C. Schroeter, and M. Volkhardt, „Robot Companion for Domestic Health Assistance: Implementation, Test and Case Study under Everyday Conditions in Private Apartments,“ in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Hamburg, Germany, 2015. |
| [3] | M. Herpers and D. Schmelz, „Generic Acceptance Test Strategy for Mobile Robots Navigation Algorithms: Applied in a Health Care Environment,“ in 10th International Conference QUATIC, Lisbon, Portugal, 2016. |
| [4] | OM Nasir, „Analysis of Requirements for Test Environment Setup for Mobile eHealth Robots“, Master's thesis, Fulda University of Applied Sciences, Department of Applied Computer Science, Fulda, 2018. |
female authors
Prof. Dr. Martine Herpers Since 2014, she has been teaching and conducting research at Fulda University of Applied Sciences in Applied Computer Science, specializing in health technology, software engineering, and gender aspects of computer science. Previously, Ms. Herpers worked for many years as a team leader in software development, test floor management, and quality management consulting in the telecommunications and automotive industries.
Robin Kirschner (BSc) She studied mechanical engineering (MSc) at the University of Chemnitz, specializing in applied mechanics and thermodynamics. Her studies on human-machine collaboration in industrial robots have been published at internationally renowned conferences.
Testing, Quality & Debugging – Our Training & Coaching
Do you want to bring yourself up to date with the latest technology?
Then find out more here MircoConsult offers training courses/seminars/workshops and individual coaching on the topics of testing, quality & debugging.
Training & coaching on the other topics in our portfolio can be found here. here.
Testing, Quality & Debug – Expertise
Valuable expertise on the topics of testing, quality & debugging is available. here Available for you to download free of charge.
You can find expertise on other topics in our portfolio here. here.
