It is not easy to build a robotic hand that can compete with a human hand, but the UK company Shadow brings us closer to that reality.
The Shadow Dexterous Hand is extremely lifelike in all of its demonstrations. It’s accurate enough to pick up an egg, screw in a lightbulb or even thread a needle, something that can be extremely difficult for human hands to do.
It has a total of 24 joints and 129 sensors, and not only do the sensors provide detailed telemetry, but they also force sensing on the joints, tactile sensing at the fingertips, temperature, and motor current and voltage sensing. The hand also has optional BioTac sensors which can be added to all fingers and allows detailed force, micro-vibration and temperature gradient sensing.
Researchers can also control the hand with a “Cyber glove”, allowing for 24 distinct human movements that mimic the user’s own hand.
The robotic hand can also make clever use of a Microsoft Kinect sensor to size up the object that it needs to grasp, and it if has seen the object before, it calls up instructions for the most practical shape to adopt to grasp it. If it hasn’t seen the object before, it runs artificial intelligence algorithms to guess the best shape to use and attempt to learn over time how to better grasp the object.
The Shadow Dexterous Hand has been integrated into other robots, such as Kuka, Denso, Willow Garage, Mitsubishi and Universal Robots, but is yet to be integrated with humans aside from the use with a glove.
It all runs on the Robot Operating System (ROS), a public library of robot programming code. The Shadow Dexterous Hand code is also Open Source so that the code can be modified, and integrated into new systems. Shadow has also provided a robot hand simulation to test out the hand in a virtual environment using ROS, so researchers can try before they buy. The Dexterous Hand is available for researchers to purchase, however the price hasn’t been made public yet.