Edinburgh Centre for Robotics (ECR) hosted its first European Robotics League for Service Robots (ERL-SR). The competition took place in Heriot-Watt University’s custom-built living laboratory. I volunteered to be part of the refereeing team. Hosting the competition was a great chance to find out more about the technical requirements, time commitments and the intensity of the competition in anticipation of a potential future entry.
ERL-SR is a robot competition in a home environment with tasks for robots to help an elderly person, named Grannie Annie with her day-to-day activities. One or two robots must perform the tasks, such as answering the door, understanding voice commands and fetching objects. The tasks require the robot to master navigation, voice recognition, visual recognition and manipulation. Navigation and mapping are already applied in the latest robot hoovers. Voice recognition is demonstrated in smart devices such as the Amazon Alexa. Accurate visual recognition for people and objects is potentially addressed by trained deep neural networks. The current available devices for object manipulation are limited to grasping and collecting empty soft drink cans/bottles, which may not provide the positive impact for service robots that the world hoped for. The path to robots delivering care to the elderly, which may involve active physical human interaction like assisting the elderly to stand up, seems to be still an open research question.
Soft manipulators can potentially overcome some object manipulation obstacles, such as handling fragile items and items with complex geometries. However, if the soft manipulator uses pneumatic actuation, then a heavy compressor and the resultant noise may not be suitable in a home environment.
The research direction for soft manipulators seems to be moving towards more dynamic and model-free approaches compared to traditional rigid kinematics and model approaches to address dynamic and non-structured environments. My own research direction focuses on an Energy Based Control approach underlined by the Thermodynamics Laws. In this example, a significant amount of energy will be required to aid a person to stand-up. The question is how to route that energy safely and efficiently. If something went wrong, where would that energy go? I think it is interesting to use a high-level abstraction approach to a precise manipulation problem.