A team of researchers from MIT announced the development of SoftZoo, a platform that helps engineers examine soft robot co-design.

The team’s work enhances algorithms with design, which shows what a robot would look like, and control, which allows for robotic motion. Overall, the structure improves how engineers generate outlines for machines in real time.

MIT doctoral student and lead researcher on the project Tsun-Hsuan Wang explained that SoftZoo helps users “understand the best strategies for robots to interact with their environments.”

The researchers’ platform includes 3D models of animals, such as bears, sharks, caterpillars, and pandas. These designs imitate soft robotics tasks, including agile turning, locomotion, and path following in various environments.

According to the research team, SoftZoo models movement that reacts to the physical features of different biomes.

This adaptability originates from a differentiable multiphysics engine that simultaneously simulates multiple aspects of a physical system, such as a caterpillar moving across a wetland or a baby seal turning on ice. This means designers could work on a robot’s brain and body at the same time to make them more specialized and aware.

This optimization reduces costly simulations needed to solve design and computational control issues and could make the robots suitable for applications like exploration and rescue missions.

However, Wang said, “Transferring from simulation to physical robot remains unsolved and requires further study.” Wang added that SoftZoo’s spatially varying stiffness, muscle models, and sensorization could not be realized with current fabrication techniques.

The team did, though, showcase the platform’s potential for human mechanics by designing a 3D arm throwing a snowball forward. Simulating more human tasks could allow soft robotics designers to evaluate arms that move, grasp, and stack objects.