Researchers have made a new robot hand that is sensitive enough to hold and manipulate not only spheres but also irregularly shaped objects. And, potentially, soft objects.
That screenshot may appear to be a gripper holding a tiny disco ball, but it is in fact a giant robot hand holding a normal-sized disco ball.
I don’t know why it’s so big, but I’m guessing it’s easier to prototype than something miniaturized. And it appears they are using the tactile finger design previously created at the same lab (Columbia Engineering).
Each finger has sensors to detect touch all over its surface. But they’re not traditional pressure sensors or switches. The sensors are made up of several light emitters (LEDs) and detectors placed all over each finger. When the outer skin touches something, underneath a sub-skin made of transparent silicone deforms. This deformation causes changes in the light which is detected and processed with software using deep learning.1P. Piacenza, K. Behrman, B. Schifferer, I. Kymissis and M. Ciocarlie, “A Sensorized Multicurved Robot Finger With Data-Driven Touch Sensing via Overlapping Light Signals,” in IEEE/ASME Transactions on Mechatronics, vol. 25, no. 5, pp. 2416-2427, Oct. 2020, doi: 10.1109/TMECH.2020.2975578.
The purely tactile approach has the added benefit that it doesn’t need a camera or other visual sensors—which in turn means it can operate in any lighting condition, even in the dark.
Now, there’s plenty of commercial and industrial robots and automation that can operate without vision. And there’s been many robot grippers that can pick up irregular and/or soft objects.
But how many can manipulate that object such as spinning it around with just the gripper?
That’s the dexterous manipulation part of this project. Columbia Engineering has successfully demonstrated the task of:
executing an arbitrarily large rotation of an unevenly shaped grasped object in hand while always maintaining the object in a stable, secure hold
Whether this is a significant advancement remains to be seen—it would be interesting to see their software adapted to a much smaller version of the hardware and actually attached as an end effector to a robotic arm, and even better with the arm attached to a mobile robot.
Here is the video from Columbia Engineering, research by Gagan Khandate, Siqi Shang, Eric Chang, Tristan Saidi, Johnson Adams and Matei Ciocarlie:
Even this little bit of manipulation is not easy. As recent as 2021, a robot grasping survey reported2Dukor, K.F., & Afonja, T. (2021). A Survey: Robot Grasping.:
Virtually all applications, from manufacturing to service to security, would benefit from robots capable of grasping any object with a wide range of shapes and sizes, from rigid to deformable, and under a variety of frictional conditions. Yet despite over 40 years of research, this problem remains unsolved.
Grippers are not the only way to achieve dexterous manipulation though. Artificial wrists could contribute as well and help with prosthetics too3Fan, H., Wei, G., & Ren, L. (2022). Prosthetic and robotic wrists comparing with the intelligently evolved human wrist: A review. Robotica, 40, 4169 – 4191.:
to achieve human-like performance, the major challenge in the research of prosthetic and robotic wrists is to achieve design balances within three pairs of contradictory factors, namely improving mobility while maintaining load capacity, enhancing flexibility while keeping stability, and empowering output capability while retaining compactness. The wrist therefore should be considered as a key research object within both robotic hand-arm systems and transradial or transhumeral prostheses.
Improvements in the abilities for grasping novel objects—as in the robot wasn’t programmed beforehand for that specific object and/or size—as well as manipulation could be useful in many fields. Especially those fields currently requiring human-level dexterity such as medical applications.
Robots don’t have to be fully autonomous of course. The company Shadow Robot, which sells robot grippers for researchers, made this little concept video of teleoperating dexterous robotic hands for the pharmaceutical industry:
- 1P. Piacenza, K. Behrman, B. Schifferer, I. Kymissis and M. Ciocarlie, “A Sensorized Multicurved Robot Finger With Data-Driven Touch Sensing via Overlapping Light Signals,” in IEEE/ASME Transactions on Mechatronics, vol. 25, no. 5, pp. 2416-2427, Oct. 2020, doi: 10.1109/TMECH.2020.2975578.
- 2Dukor, K.F., & Afonja, T. (2021). A Survey: Robot Grasping.
- 3Fan, H., Wei, G., & Ren, L. (2022). Prosthetic and robotic wrists comparing with the intelligently evolved human wrist: A review. Robotica, 40, 4169 – 4191.