Researchers have developed a system for soft robot co-design, which entails the collaborative search for and optimization of robot design. It includes the robot's shape, where the muscle is placed in the robot's body, and the degree to which different parts of the robot's body are rigid or soft.

"Robots are not going to replace humans. Instead, they are going to make their jobs much more humane. Difficult, demeaning, demanding, dangerous, dull – these jobs robots will be taking." - Sabine Hauert.

What is soft robotics?

Soft robotics mimics living things mechanically. Engineers in "soft robotics" have been busy creating various representations of flexible machines with applications in exploration, locomotion, rehabilitation, and space since the word was adopted in 2008. For example, animal behaviour in the wild can serve as a model for how to move. Taking this a step further, a group of MIT researchers built SoftZoo, a bio-inspired platform that allows engineers to examine soft robot co-design. Improvements to the automatic generation of machine blueprints are made possible by the framework's optimization of algorithms for design (which determines the robot's appearance) and control (which permits robotic mobility).

3D models

The platform takes a more naturalistic approach by including 3-D models of creatures, including panda bears, fish, sharks, and caterpillars, as designs that may emulate soft robotics activities like locomotion, nimble turning, and path following in various settings. The platform exemplifies the performance trade-offs of several methods in multiple environments, including snow, desert, clay, and water.

Engine Features

SoftZoo is more thorough than other platforms that currently simulate design and control. A differentiable multiphysics engine gives the framework flexibility, allowing for the simultaneous simulation of multiple features of a physical system, like a baby seal spinning on ice or a caterpillar creeping across a swamp. The differentiability of the engine improves co-design by minimizing the number of time-consuming and costly simulations needed to address computational control and design issues. In addition, it allows for more complex, tailored algorithms to be used to design and control soft robots.

Furthermore, the system's ability to simulate interactions with diverse terrain demonstrates the importance of morphology, studying organisms' shapes, sizes, and forms. Some biological structures are better suited to a given environment than others, much like comparing the designs of different machines that do the same function. 

What is the upgrade?

In the past, robots had difficulty navigating through crowded spaces since their bodies couldn't adapt to certain conditions. However, with SoftZoo, designers may simultaneously work on the robot's brain and body, leading to co-optimized intelligent and specialized land and sea robots. As a result, the robots' use in search-and-rescue operations and geographical exploration would improve with higher behavioural and morphological intelligence levels. For example, if a person went missing during a flood, an optimized robot using the techniques shown in the SotftZoo platform could more easily navigate the water and find the missing person.

Conclusion

Getting data from a virtual robot to a real one is still a problem, adds Wang, and it needs more research. As the authors put it, "The muscle models, spatially varying stiffness, and sensorization in SoftZoo cannot be straightforwardly realized with current fabrication techniques, so we are working on these challenges."

Because it may be used to evaluate robotic control, the platform's creators are considering future uses in human mechanics, such as manipulation. Wang's group built a three-dimensional arm that could throw a snowball ahead to show off the technology's potential. To evaluate soft robotic arms that can grasp, move, and stack objects, designers in the field of soft robotics would benefit from the platform's ability to simulate more human-like tasks.

Image source: Unsplash