What would you do if you have a robot that can easily catch fast moving objects?
The researchers from the Learning Algorithms and Systems Laboratory at EPFL in Switzerland are developing a way to transfer information from a human to a robot, enabling it to catch objects thrown at it. The most impressive thing here is that the robot is able to catch objects that are both statistically and dynamically unbalanced. A baseball, a balanced object, is relatively easy to model as it will follow a predictable trajectory depending on its speed, direction and gravity. On the other hand, a half-filled water bottle or hammer would be very challenging as they would tumble or spin erratically due to the inconsistency of their center of gravity.
The research is being conducted with the use of a Kuka LBR iiwa robot. By manually guiding the robotic arm to catch the objects, the Kuka robot is able to learn by demonstration. It then constructs a mathematical model that would allow it to react on its own in real time.
This research would open a wide range of application, where fast reaction time in unpredictable environments is necessary. It may be useful to protect people from incoming fast moving objects or other dangerous situations.
Source: IEEE Spectrum