Research

  1. Dynamic handover: In the future, robots will help humans in everyday tasks, such as, handing over a hammer in a workshop, or bringing a cup of water to an elderly. One of the most important physical interaction channel between humans and robots is object transfer. However, robots need to adapt their behavior to several environmental features. For example, they have to know how to hand over a glass of water at home, hand over a hammer to a worker, or just a pill to a patient. Moreover, robots need to adapt to human features, such as, age, height, human preferences etc. In our research we aim at learning and adapting to individual humans using their high level feedback, e.g. “I preferred this handover over the previous one”, or “I think this was a perfect, 10/10 handover”. By using this adaptation technique, future service robots will not have to be programmed perfectly upon delivery, but they will be able to adapt autonomously to the need of human partners. For the dynamic handover exhibit, the Kuka LBR iiwa arm will be holding an object like a water bottle. Visitors can walk past the Kuka LBR iiwa arm and attempt to grab the object. This shows visitors the ease of humans interacting with robots in the task of object handover.
  2. Soft robotic gripper
  3. The macro/mini robot framework is a cooperative system where a large industrial type robot manipulator (macro) carries a smaller specialized robot (mini). The current experimental setup uses a serial-link KUKA KR60 industrial robot with a prototype parallel-link robot built by NUS graduate students. The macro enables the mini to work within a very large workspace while the mini is able to provide fast and light position, force, or impedence control. Present research is aimed at replicating the human skills used in deburring metal work pieces and investigating other light-force interaction tasks, such as polishing and grinding.