Creating a robot from scratch is a daunting task. For those without engineering experience, designing a robot might feel entirely out of reach. Even with the experience, it could take years of training, testing, and optimizing before a functional product hits the ground running.

Cynthia Sung — an associate professor in the Department of Mechanical Engineering and Applied Mechanics at the University of Pennsylvania — attempts to make robot construction more accessible by reimagining core robotics issues into engineering- and biology-based problems. On September 20, she showed U of T students how as part of the university’s Robotics Institute Seminar series.

From planning the path of an airplane to designing a robotic arm

You wouldn’t think that paper folding could create a robot. Yet, the origami-inspired robotic designs — based on traditional paper-folding art — have the advantage of being lightweight and versatile, offering a unique solution to the classic issue of too much weight for the intended motion or speed. Sung uses thin sheets of plastic, folded like origami, to make the bodies of these new robots.

However, if we were to design an origami bot, we would have a hard time doing it with current computation methods, as it would be a slower process and leave more room for error. Sung’s Robotics Lab has found a way to automate the design process from start to finish. She showed that designing a robotic arm can be simplified to a path-planning problem. 

The engineering path-planning problem is assessing the most optimal way for something to reach a certain goal. It’s often programmed into self-driving cars. Finding a time-optimal path for a plane from point A to point B in three-dimensional space can be solved by connecting segments of straight lines and curvature movements, which are both automatable calculations.

Designing a robotic arm path, for example, can then make similar use of the optimization problem: where does there need to be a straight rigid link — joints between parts that do not move — and where can we put a rotational joint — one that can rotate freely? That is, for a robotic arm’s movement to be optimal, where should the various types of joints be to put the least amount of strain on each one?

The automation tremendously streamlined the design process. Using this approach in their algorithm, they were able to construct a four-legged robot that could roam around the Philadelphia sidewalk within a day.

Biology that inspires physical mechanics

Designing an energy-efficient robot is just as important as its movement capability. Many animals for instance have energy-efficient systems through springs and energy storage. To reduce the demands of their muscles, many vertebrates turn to tendons and ligaments to store the energy generated by their contraction and relaxation as they move. 

As it turns out, geometry can affect the stiffness of a system and therefore, how the energy is distributed. 

Sung’s lab successfully modified their algorithm to leverage this insight, enabling energy conservation through the robot’s structural properties. With the algorithm taking inspiration from the geometry of a squirrel’s paw, their four-legged bot could be modified to grasp onto fake branches. 

“This was really interesting from a point of view of robot design,” said Sung, “but how does this robot start to interact with [its] environment?”

Walking on terrestrial or lunar sand?

Initial studies performed on robots navigating on sand show that they can gather information about their surroundings through ways like measuring the force exerted by the ground onto the robotic foot. Roboticists are looking to implement this into their next-generation robots’ computational design, to make them adaptable to different types of environments.

This has piqued the interest of NASA — who is funding the current work — in the hopes of sending some of these robots onto the moon to navigate alien terrains. Increasing the agility of these robots will ultimately help in answering a range of societal needs, including space exploration, surgery, and prosthetics. Breaking down the robot-constructing process is one step — and an exciting path to do so is being paved.