Climbing Robot (with Claws!)

I'm Paul Naidan. I am a PhD student in the Robot Mechanics Lab at Carnegie Mellon University. Kind of the goal of our lab is to help get robots out into the real world, not just kind of controlled lab environments. The biggest challenges for getting robots in the real world is handling contact, so kind of where the robot meets the environment. But then kind of the focus of my project has been on kind of adhesive contact where the robot needs to create forces to hold itself onto a vertical surface. There's a lot of different adhesion mechanisms like magnetic or suction. The one used by kind of my robot, Loris, is microspines, which is basically lots of sharp little hooks that catch on all the kind of small microscopic bumps and pits in the rock surface. The name Loris stands for Lightweight Observation Robot for Irregular Slopes, and the kind of the origin of the name comes from the Loris, which is kind of a small climbing primate. It's very small. It has a kind of very slow, cautious manner of climbing that really matches how this robot works in practice. So here Loris is attached to some of the rocks outside the conference venue. You can kind of see the way the grippers work. They have two sets of spines at a 90 degree angle, and this design helps them support forces that are going at different angles, whereas normally microspines are kind of designed to apply force in a straight line. And you can kind of see the robot is applying forces between the two different sides, and that produces some extra preload that gives it more adhesion. Hopefully a mix of like bio-inspiration plus just kind of engineering design where the kind of the idea of directed inward forces between the feet originally came from other labs studying lizards and how like geckos and how they climb. Microspines were originally based on kind of cockroach spines, not geckos. The tail provides a lot of leverage to help keep the robot from falling backwards as it's climbing, and so by pushing with the tail you reduce the amount of force you need to adhere it to the wall with using your grip groups. The kind of key idea with Loris is using under actuation to reduce the amount of mass needed for kind of motors and so on to kind of keep the robot smaller and lighter. The grippers are fully passive, so instead of having a motor at them that kind of closes them around wherever it's grasping, the robot instead uses kind of directed inward grasping forces between different feet where they can kind of generate internal forces between each other but without adding any new motors because you're just using the ones you already have for your legs. Microspines work really well on kind of rough rocky surfaces where there's kind of all those small bumps and pits for them to catch on. So kind of the main environment we've been focusing on is kind of space applications, so kind of steep cliff faces or kind of subsurface lava tubes and caves on like the moon or Mars. It's like wheeled rovers have like done a lot of science on Mars but there's a lot of kind of areas we can't access so kind of if you have a kind of steep cliff you'll have access to all those kind of sedimentary layers. You can kind of get a geological history almost by kind of having being able to observe those or ideally even like collect samples and bring them down to a larger rover that actually has all the science equipment on board. We're currently working on kind of the autonomous path planning and foothold selection. So the depth camera is used to kind of build a map of the terrain and then we're working on using kind of sample based planners to find a kind of a safe route using that map as well as select footholds that are going to be the kind of right size and shape to really adhere to at the grippers.