Skin-like multi-modal sensing devices for dexterous robotic hands
- Skin-like sensing would be useful for many applications, ranging from human-friendly robots to prosthetics. However, while tactile sensors have been studied since the 1970s, they remain relatively little used in applications. The main challenges are practical: there is a need for ``electronic skins'' that are sensitive, flexible and even stretchable, and robust enough to cover the surfaces of arms and hands with many sensing elements. The devices should have high sensitivity and be multi-modal, i.e., able to report changes in normal and shear stress, as well as temperature and proximity. The sensor skins should also be low cost and robust. The goal of my thesis work is to provide skin-like cutaneous sensing ability to robotic hands. The chapters in this thesis include manufacturing techniques and materials for flexible, low-cost, lightweight, sensitive, robust, and multi-modal sensory skins. The sensors include multiaxial force/torque elements, proximity sensors, and temperature sensors. Briefly, the work presented here provides the following contributions in support of cutaneous sensing: fabrication (patterning and bonding), material (dielectric elastomer), sensing devices (multi-axial and multi-modal sensors, and sensor network), integration (soft robotic skin for manipulation with robotic hands). The first contribution concerns manufacturing techniques to fabricate sensing devices in thin films. The sensors and electrodes are created using an ultraviolet laser to ablate and cut patterns on metalized plastic film. With this computer-aided subtraction-based fabrication process, the sensors and their arrangement are easy to customize for different applications. A titanium-induced bonding technique is also introduced and shown to achieve strong bonding between an electrode and a dielectric layer. The rendered strong adhesion on the interface of the device component allows a robust capacitive sensor to withstand significant shear and rotational loads, which increases the sensor's dynamic range. Next, a material is presented to enhance the sensitivity of capacitive sensing devices. This material is afforded by a combination of a titanium-based solution and a stretchable elastomer. We observed that the titanium oxo network and silicone elastomer matrix composite have the potential to provide stretchability and adhesion for the dielectric, for a robust and sensitive device. Using the developed manufacturing techniques and materials, we next developed multi-modal sensing devices inspired by the human cutaneous sensing system. We explored capacitive multi-axial sensor designs that conform to curved surfaces, allowing them to wrap around the back and sides of a robotic hand. Each taxel measures a combination of normal, shear, and torsional stresses. With active shielding and a microstructured porous dielectric material, the sensor has a desirable combination of wide dynamic range with high resolution (i.e., 0.5 to 500 kPa in the normal direction), relative immunity to electromagnetic noise, and the ability to handle wet and slippery materials, such as tofu or Jell-O. By dynamically changing the patterns and combinations of electrodes sampled, our developed sensor can provide dynamic as well as low frequency tactile information, even when scaled to large areas. Empirical results indicate that our sensor can detect changes in grasp force and events such as making or breaking contact and the onset of linear or torsional sliding. An additional development consists of a low-cost, stretchable Kirigami sensor network for soft robotic devices. Soft robotic hands can facilitate human-robot interaction by allowing robots to grasp a wide range of objects safely and gently. However, their performance has been hampered by a lack of suitable sensors. To address this, we developed a multi-modal sensor network integrated with a soft robotic hand. The manufacturing approach uses UV laser ablation to create sensor patterns and UV laser cutting for stretchable interconnections in a Kirigami pattern. Temperature and proximity sensors are combined into a single network. We evaluated both the interconnects and sensors by measuring the impedance change to external stimuli and observed that sensor readings are not substantially affected by stretching the interconnects. We tested several interaction scenarios, including identifying hot and cold water bottles, a warm burrito for food handling, and a warm baby-like doll for future medical applications with the sensor sheet wrapped around a soft robotic gripper. Our demonstrations showed that robotic hands can have cutaneous sensing that provides robust and reliable multi-modal information. In summary, we have fabricated sensing devices and networks using laser ablation and reliable bonding techniques. The multi-axial and multi-modal cutaneous sensing devices are demonstrated on two robotic platforms. The first is a gripper that manipulates a variety of fragile and slippery foods. The second is a soft robotic hand that measures temperature while maintaining gentle contact. Collectively, our experimental results show that feedback from the tactile sensors enables robots to identify contact stimuli in support of an improved understanding of objects and the environment with which they are interacting
|Type of resource
|electronic resource; remote; computer; online resource
|1 online resource
|Cutkosky, Mark R
|Cutkosky, Mark R
|Kenny, Thomas William
|Degree committee member
|Kenny, Thomas William
|Stanford University, Department of Mechanical Engineering.
|Statement of responsibility
|Submitted to the Department of Mechanical Engineering
|Thesis Ph.D. Stanford University 2020
- © 2020 by Jooyeun Ham
- This work is licensed under a Creative Commons Attribution 3.0 Unported license (CC BY).
Also listed in
Loading usage metrics...