Robotic Finger Sensing Advances Human Interaction

In this case, the digit is a finger. Advanced robots designed to interact on a personal level with humans will need hands that can sense a variety of interactions. Several sensing techniques could be employed to achieve the right level of dexterity and avoid excessive force — some simultaneously and far less for specific applications. Getting the right sensing solution has been the goal of several universities.

Sensing in the Biomechatronics Lab at UCLA

Researchers at the University of California, Los Angeles (UCLA) have given a robotic finger the ability to detect and react to physical contact by sensing multiple touch modalities. The finger is equipped to measure vibration and internal fluid pressure, has an external material that is soft to the touch, and even measures temperature. Electrodes and a pressure sensor inside the mechanical digit provide outputs that can be displayed during design and development.

Robotic finger with integrated sensors. (Image: Ancient Aliens)

Exposed electrodes in digital finger. (Image: Ancient Aliens)

Shown without the external covering, the dots are the electrodes in a rigid core surrounded by an elastic, liquid-filled skin. The electrodes indicate the amount of deforming that occurs with interaction when the finger pad is squished. This deformation allows the digit to distinguish between a soft material, such as foam, and a hard material, such as glass, since they have different stiffness or rigidity values.

Measured deformation. (Image: Ancient Aliens)

Another group of researchers at UCLA developed a glove-based system to perform simultaneous hand pose and force sensing in real-time. Their goal was to measure and collect human hand data during fine manipulative actions so the system could visualize hand actions. The sensory glove captures the data of finger poses and hand poses, as well as forces on the palm and each phalanx, using a network of 15 inertial measurement units (IMUs) to measure the rotations between the individual phalanxes. With this information, a hand pose is reconstructed using forward kinematics. Contact forces on the palm and each phalanx are measured by six customized force sensors made from a flexible piezoresistive material.

MIT researchers’ camera-based sensing

Researchers from several departments at the Massachusetts Institute of Technology (MIT) collaborated to create a camera-based touch sensor in an object shaped like a human finger. Called the GelSight Svelte, the sensor has two mirrors to reflect and refract light with the camera, located in the sensor’s base, capturing images along the entire finger’s length. The camera views the back of the skin from the inside of the glove. This design provides high-resolution tactile sensing over a large area.

The finger-shaped sensor has flexible “bones” similar to human phalanges. The bones, which have two sets of LEDs for illumination attached, are encased in flexible silicone gel skin. By measuring how the bone bends when the finger touches an object, the researchers can estimate the force being placed on the sensor.

A robotic hand was designed to employ three sensors to perform multiple grasps. Using the entire sensing area of the three fingers, the hand can execute a pinch grasp, a lateral pinch grasp, and a power grasp.