Morgan Pope is a PhD student investigating robots that live at the boundary of airborne and surface locomotion at Stanford’s Biomimetics and Dexterous Manipulation Lab. He wrote about SCAMP, a flying and perching robot, for Automaton earlier this year.
A disaster site. A rainforest. A battlefield. These places have something in common: we have a need to understand what’s going on where established infrastructure can’t give us good data. Advances in computation, fabrication, and materials over the last half-century have resulted in small, cheap, and lightweight sensors that can provide us with these data; now the task is to find ways to deploy such sensors rapidly and effectively.
One way to do this is with small, agile aerial vehicles like quadrotors. Quadrotors are becoming affordable, ubiquitous platforms that can fly quickly over rugged terrain to collect critical data. There’s a catch, though: most small (less than 1 meter in diameter) quadrotors can only stay in the air for tens of minutes at a time, and this limited endurance makes some missions unachievable. However, if the goal is to collect data from a fixed vantage point, there is an alternative to hovering in place that might extend mission life from minutes to days or even longer: perching.
Perching allows a quadrotor to shut down its power-hungry motors and let its sensors get to work acquiring data over an extended period of time, tracking parameters like the stability of a building after an earthquake, the nocturnal activity of a jaguar, or enemy troop movements. While perched, the quadrotor can also happily continue to operate in weather conditions that would make flying impossible. At Stanford’s Biomimetics and Dexterous Manipulation Laboratory, we have been working on perching with the goal of making landing on a wall as easy as landing on the ground. By adding a few grams of structure and mechanism to an off-the-shelf commercial quadrotor, we are now able to perch on both vertical and inverted surfaces without using any special firmware or flying techniques. While it’s still not as foolproof as landing on a level surface, we are closer than ever to making perching accessible outside of a research environment.
This solution originated when Hao Jiang, a graduate student in our lab, took a closer look at another recent result from our group, a perching/climbing robot I built called SCAMP, which uses a climbing mechanism mounted on top of the robot. Mounting the mechanism in this way meant that the thrust from the robot’s rotors actively push the robot onto the wall during perching, assisting with the maneuver. By first impacting the wall with a rigid carbon fiber “tail,” SCAMP creates a stable pivot point to transition from level flight to pitching up parallel with the surface.
SCAMP also used an on-board computational routine to detect impact and actively pitch the quadrotor into the surface. However, as Hao and I investigated this behavior, we realized that this active maneuver wasn’t strictly necessary. If Hao could outfit the machine with a gripping system that was centrally located between all four rotors and hit the wall with a reasonable incoming velocity, the physics of the system should consistently lead to good contact with the surface.
To test out this new perching technique, Hao built a gripping system that, while unable to climb like SCAMP, has the advantage of being centrally located and capable of attaching to an inverted surface like a ceiling (a useful feature when you want to get your robot out of the rain). He then distilled the mechanical design of SCAMP into a simple tail structure that helps the robot engage with the wall. After some trial and error, Hao found that he could perch successfully by simply flying the quadrotor straight into the wall: as long as the quadrotor is moving at a reasonable pace and is pitched forward (usually the case when flying forward) and squared up with the target surface, the rotors reliably bring the mechanism into contact to engage the wall using an opposed grip.
This opposed grip strategy works by dragging two sets of microspines—hardened steel spikes on a special suspension— along a surface in opposite directions. The spines catch against tiny bumps and pits on the surface and hang on using friction. Pulling the opposed sets of spines against each other produces a tight grip that lets the quadrotor land not only on vertical walls, but also on angled or overhanging surfaces. As Hao explains it, “the opposed-grip strategy for microspines is just like a human hand grasping a bottle of water, except that while humans require some macroscopic curvature to get our fingers around both sides of an object, the microspines can go deep into the micro-features of a rough surface and latch on those tiny bumps and pits.” When the frequency of small bumps and pits is high, as with stucco or cinderblock, the grip is more reliable than on surfaces like polished concrete. Soon after perching on the wall, Hao started flying the quadrotor up into ceilings, and found that the inverted surfaces were often even easier to perch on because the quadrotor was already aligned to the surface in its normal flight configuration.
“The opposed-grip strategy for microspines is just like a human hand grasping a bottle of water, except that while humans require some macroscopic curvature to get our fingers around both sides of an object, the microspines can go deep into the micro features of a rough surface and latch on those tiny bumps and pits.”
Once perched securely on a wall, Hao had to figure out how to enable the quadrotor to release and fly off. On an inverted surface, it was easy: a servo releases the spines and gravity does the rest. On a vertical surface, it took a little more thought to make sure the quadrotor would rotate away from the wall properly, since the thrust from the rotors tends to keep the robot tucked tightly against the surface. As Hao explains, the solution was to add some spines on the tail: “During take-off, as the mechanism is released and the quadrotor starts to fall, the microspines on the tail catch on bumps and pits [on the surface] again. The quadrotor then pivots on the tail spines backward away from the surface, and can fly away.”
Speaking about challenges and future ways to address them, Hao says that, “even if the perching strategy is robust, the quadrotor can still fail to perch due to improper choices of target surfaces. While we have achieved robust perching failure detection and recovery for indoor environments, we will investigate failure recoveries for outdoor applications, possibly with wind disturbances and surface uncertainties.” We are also interested in pursuing new gripping strategies and possibly combining microspines with dry adhesives to stick to a larger range of surfaces. We’re excited by the recent advances in perching and are hoping to refine our approach even further, so that people can start putting sensors where they need them the most.