Image credit: Joseph DelPreto/MIT CSAIL
Companies use robots and robotic technologies to make their manufacturing floors safer and more efficient. Many of these robot helpers are used for high-volume, repetitive tasks that can cause injuries to human workers, but they are often sharp and unwieldy, not exactly conducive to collaborative work.
Soft robotics companies are developing flexible materials that will extend the capabilities of robots on the factory floor. New technologies will allow robots to extend, contract, or bend in response to simple control inputs, enabling them to grasp and manipulate objects more easily. In some cases, the robots will have the ability to lift 1,000 times their weight. That will alter how robots can be used on the manufacturing floor.
Researchers at Columbia Engineering have developed a soft robotic muscle that is 3D printed. The synthetic muscle is three times stronger than human muscles and costs pennies to produce. It uses a lightweight, self-contained, soft actuator made of a silicone elastomer and ethanol. When electrically activated, the ethanol heats up and expands, moving similarly to a muscle.
In tests, the researchers created “bicep” and “tricep” muscles that were attached to a skeleton. Activating the biceps raises the forearm. Rather than waiting for the biceps to cool off to revert the arm to its home position, the team found it was faster to engage the triceps to bring the arm down. The synthetic muscle can lift 1000 times its own weight.
Image credit: Aslan Miriyev/Columbia Engineering
The technology will allow for high-tech advancements in robotics. The material can push, pull, bend, twist, and lift weight. It will allow “hard” robots to have more human-like qualities and capabilities. Soft robots have flexible shapes, allowing them to fit in small spaces, adjust to different climates, and imitate or function as part of the human body. One drawback to the Columbia Engineering synthetic muscle is that movements can be slow.
Robots Lend a Hand
A quicker, and more cutting-edge alternative is being developed in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). RoboRaise is designed to work in collaboration with a human to lift objects. When two humans lift an object such as a couch, they adjust to each other’s movements. RoboRaise operates in a similar fashion using artificial intelligence to monitor and mimic a person’s muscle activity.
The solution is based on using nonverbal cues that encode instructions on how to coordinate movement. The human wears electromyography (EMG) sensors on their biceps and triceps. The RoboRaise algorithms monitor muscle activity to detect changes to the person’s arm level and vertical hand gestures, which enable more motor control. In tests with a human, the system was able to pick up and assemble mock aircraft components with about 70 percent accuracy.
Using wearable technology circumvents potential problems with speech, line-of-sight, and ambient noise. The sensors allow humans to control the robot’s movements though muscle signals. First-time users need to train the robot by tensing and relaxing their muscles a few times and lifting a light weight to various heights. The robot’s deep learning capabilities detect and respond to the gestures. The more the robot is used with humans, the better RoboRaise can adjust to nuances in movement.
In the future, the CSAIL team will work to add additional sensors or incorporate more muscle groups to bring greater utility to RoboRaise. The team expects the device to be used initially in the manufacturing and construction industries. The project is partially funded by The Boeing Company.
CSAIL researchers are also exploring the possibility of the robot sensing fatigue in the human muscle. RoboRaise could then adapt its lifting capabilities to prevent strain on the human body. That could help reduce worker injuries on the production floor.
Robot assistants are increasingly integral to production and manufacturing environments. However, few are designed to work alongside or interact with humans directly. With these new technologies, it won’t be long before humans and machines operate in tandem to achieve a shared goal.