A world's most advanced robotic hand
Google Deep Mind is taking a new approach to advance the field of Artificial Intelligence in robotics by introducing a dexterous and durable three-fingered robotic hand, developed by UK-based Shadow robot, this robotic hand has been meticulously designed to match the strength and precision of movement of the human hand.
The new series.
The Shadow dextrous hand series from the company known for its five-fingered robotic hands is hailed as the most advanced in the world, offering exceptional precision that allows the manipulation of tools and objects with unprecedented dexterity, the Shadow robot website tells us showcases this technology and not only opens new possibilities for research in robotics and Artificial Intelligence, but also improves existing technologies by raising the standard of excellence in the field.
The report published in New Scientist reveals that this new 3-fingered robotic hand has been designed to withstand repeated abuse, such as direct impacts to its fingers from various directions or being crushed by objects such as a hammer, with a fast transition speed and open to closed in just 500 milliseconds and a fingertip clamping capability with up to 10 newtons of force.
This robotic hand demonstrates an exceptional ability to perform a variety of tasks, while the existing Shadow dextrous hand series seeks to replicate the kinematics and dexterity of the human hand. The new three-finger model incorporates significant advances, weighing approximately 4 kg and 1kg per finger is close in weight to the Shadow dextrous hand and forearm providing a balanced structure for versatile operations with the power to handle objects up to 11kg.
The best materials.
The hand is constructed from a combination of durable metals and plastics including aluminum, brass, acetyl polycarbonate and polyurethane, ensuring strength and flexibility for a variety of tasks with efficiency and precision, equipped with 20 motors built-in temperature, voltage and current sensors, as well as extensometers in the tendons and 26 Ho effect sensors to meticulously track joint positions with 0.2º resolution, the palm features an IMU that houses a three-axis gyroscope and an accelerometer for comprehensive movement tracking, In addition, touch pressure sensors on the fingertips improve tactile sensitivity and enrich data collection.
Mutual Learning.
Robotics companies can teach Shadow dexters hand to perform goal-oriented tasks and control tangible elements naturally, Researchers have trained robotic hands in simulation using neural networks and trial and error method, i.e. reinforcement learning The data is then sent to the robot's hand allowing it to perform the necessary task in real time and in a real environment.
With a position control mesh operating at 1 k Hertz, this robotic hand offers fast and synchronized processing of control variables enabling fast and precise manipulations by providing a wealth of data and complete sensing capabilities, Shadow robot trains researchers to explore new frontiers in robotics and artificial intelligence, promoting innovation and advancement in the field.
