Bipedal walking has worked out pretty well for humans. I guess. We’re kind of stuck with it until someone comes up with something better. And the really frustrating part is that all kinds of animals have already come up with better ways of getting around: specifically, birds and insects, who use wings to fly as well as legs and feet to walk. This multimodality makes birds and insects inherently versatile and adaptable, which is why you can find them doing quite well just about everywhere.
Some of the most versatile and adaptable robots also exhibit multimodal characteristics: they can fly and climb, or jump and glide, or even fly and swim. But flying and walking seems to be by far the most useful combination, as evidenced by the variety of animals that can do it, and researchers at the University of Pennsylvania’s GRASP Laboratory have designed a new robot called Picobug that can fly, walk, and even (soon) grab on to stuff.
This lovechild of a quadrotor and a UC Berkeley’s DASH robot displays agility in both terrestrial and aerial operations. The flying bit is straightforward: most of the robot consists of a 22-gram Dragonfly picoquadrotor with a custom autopilot. Underneath is an eight-footed, symmetrical adaptation of DASH, which has the advantage of making sure that the robot is supported by four feet at all times with minimal deformation. The motor that drives these feet can only get them going forwards or backwards, but the Dragonfly’s rotors can yaw the robot to steer it left and right.
In terms of performance, Picobug can fly with a top speed of 6 m/s, and crawl at up to 0.16 m/s on a flat surface. It weighs a total of 30 grams, which is a whole lot of not much: it’s about a third battery, a third motors and props, and the rest is the crawler, frame, and electronics. The big question is, as always, battery life: how much of a difference does the crawling actually make? The simple answer is that the robot uses 10.6 watts while hovering, and 0.6 watt while crawling, resulting in a flight time of 10 minutes and a crawl time of 45 minutes. But since flying is much faster than crawling, the complicated answer involves calculating cost of transport, a dimensionless measurement of how much energy the robot expends to move itself a given distance. The cost of transport while flying works out to be about 36, but just 14 while crawling, meaning that the robot is more than twice as efficient while toddling along on the ground.
It’s important to keep in mind that the whole point of having a multimodal robot, though, is that you can use different modes of locomotion depending on where your robot is and what it’s trying to do: you’re not just restricted to one thing. As the video shows, you can hop into the air to jump over obstacles or rough terrain, while crawling is best for delicate positioning on the ground. And for traveling through tight spaces, using flying and crawling at the same time might be the most effective technique.
The addition of a gripper to the robot (even though it seems to be a bit of a work in progress) is what has the potential to make this robot even more amazing, as the researchers speculate in an upcoming ICRA paper:
The Picobug can easily pick up small objects (on the order of 6mm and 1-2g), although a larger and stronger gripper should enable it to pick up even larger objects. With its multi-modal capabilities, the Picobug can pick up an object while in crawler mode, deliver it to its destination by air, and then return to crawler mode to deposit the object. These capabilities make the flying monkey a powerful tool for object retrieval/delivery and, when coordinated in swarms, for the construction and disassembly of structures.
Furthermore, since the gripper is not an integral part of the Picobug’s structure, it can be replaced by mechanisms with other functions, such as a mating device that allows it to couple with another robot or to latch onto a wall or branch.
While the three capabilities enabled in the Picobug are sufficient to complete a variety of tasks as listed above, we envision the next generation of such devices to include other abilities, such as cutting /milling/machining, heating/cooling, deposition of glue, etc to facilitate a wider set of applications. Future work must draw from research in swarms as such functionality will only be achieved through the coordination and cooperation between groups of devices with different sets of abilities.
Imagine a huge swarm of these robots, each with a different attachment, all working together to accomplish tasks that no one robot could do by itself. It’s quite a vision, and we like it.
For more details, we spoke with Yash Mulgaonkar, first author on the paper on Picobug and PhD student at GRASP Lab:
IEEE Spectrum: How does your multimodal robot compare with other small robots that can walk or drive as well as fly?
Yash Mulgaonkar: The Picobug originates from a new innovative category of robotics called “Printable Robotics” that uses rapid fabrication techniques for its construction. The quadrotor itself is a conventional FR4 printed circuit board integrating all the essential electronics as well as serving as the primary structural component of the Picobug.
The crawler is designed by our collaborators at MIT using the PopupCAD technique developed at Harvard. It is manufactured out of laminated sheets of the same FR4 material with embedded plastic hinges. The crawler starts off as a simple 2D multi layer laminated FR4 sheet and is simply folded into shape, drawing inspiration from origami.
The main contribution of the Picobug is the nonlinear controller software and the use of minimal hardware, consisting of only one motor for the crawler to move forward and combining the control from the quadrotor for turning on the ground. Typically most multi-modal robots carry redundant hardware e.g. extra motors for moving and turning on the ground in addition to those needed for flight.
What are some of the challenges that you had to solve while designing this robot?
Building a multi-modal robot of this scale was nothing short of a challenge. The Picobug has an all-up weight of 30 grams, making it one of the smallest multi-modal robots. The autopilot for the Picobug was custom designed to integrate all the necessary sensing, processing, communication, and motor drivers into a 30 mm x 30 mm package weighing 4.8 grams, which is less than a U.S. 25-cent coin. Miniaturizing an autopilot to this scale without any loss in functionality required placing over 200 electronic components in a tight footprint without violating any component design criteria.
The crawler is also designed to be as light as possible, weighing only 4 grams, and has four articulated legs with a complicated linkage mechanism driven by a single motor. Designing such a complicated mechanism from a laminated 2D sheet would not have been possible without PopupCAD.
Can you give examples of some situations in which the Picobug is uniquely capable?
Because of its small size, the Picobug can explore environments that other robots would not be able to access. Its ability to switch between flying and crawling modes gives it greater flexibility in its motion planning, not only because it can reach areas that other robots can’t, but also because it can extend its effective mission life by optimizing its path. Finally, its low-cost design makes it ideal for applications where it is likely that the robot will not be able to perform its task and then return home.
An immediate scenario is the exploration of nuclear reactors or other facilities where there could be debris and the robot is not expected to be recovered: the Picobug can navigate through narrow spaces such as holes on walls, debris from fallen structures or narrow pipes or air ducts either by flying, walking, or a combination of both. It can be used for video surveillance, or for deploying small illumination modules, audio sensors, or other devices needed for in-situ analysis.
You suggest that it might be possible to couple several Picobug robots together. What kinds of things would that allow you to do?
Multiple Picobugs could join together to form a single larger robot to perform a task that one robot alone would not be able to do, such as manipulation or transportation of heavy payloads or other larger robots, just like fire ants do in nature. When we dream up possible scenarios involving many Picobugs, one possibility is that a large swarm of the robots could one day form a large bridge or some other large structure—the sky is literally the limit!
“The Picobug: a Mesoscale Robot that can Run, Fly, and Grasp,” by Yash Mulgaonkar, Brandon Araki, Je-sung Koh, Luis Guerrero-Bonilla, Daniel M. Aukes, Anurag Makineni, Michael T. Tolley, Daniela Rus, Robert J. Wood, and Vijay Kumar, will be presented at ICRA 2016 in Stockholm in May.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.