LEONARDO: The Hopping, Flying Bipedal Robot

LEONARDO, a hybrid drone and bipedal robot

We appear to have a new entry atop the “Robots That Creep Us Out” leader board: meet LEONARDO, the combination quadcopter/bipedal robot.

LEONARDO, a somewhat tortured name derived from “LEgs ONboARD drOne,” is actually just what it appears to be: a quadcopter with a set of legs. It comes to us from Caltech’s Center for Autonomous Systems and Technologies, and the video below makes it easy to see what kind of advantages a kinematic mash-up like this would offer. LEO combines walking and flying to achieve a kind of locomotion that looks completely alien, kind of a bouncy, tip-toeing step that really looks like someone just learning how to walk in high heels. The upper drone aspect of LEO provides a lot of the stabilization needed for walking; the thrust from the rotors is where that bouncy compliance comes from. But the rotors can also instantly ramp up the thrust so LEO can fly over obstacles, like stairs. It’s also pretty good at slacklining and skateboarding, too.

It’s easy to see how LEO’s multimodal locomotion system solves — or more accurately, avoids — a number of the problems real-world bipedal robots are going to experience. For now, LEO is pretty small — only about 30″ (76 cm) tall. And it’s rather lightly constructed, as one would expect for something that needs to fly occasionally. But it’s easy to see how something like this could be scaled up, at least to a point. And LEO’s stabilization system might be just what its drunk-walking cousin needs.

Thanks to [qes] for the tip.

[via Tech Xplore]

22 thoughts on “LEONARDO: The Hopping, Flying Bipedal Robot

    1. Walking robot mode might actually work out to be quite efficient vs other mechanical bipeds – by using the thrusters to aid the gait the legs are much lighter and simpler than a true biped, so you are trading lighter less energy to move legs for having a small balancing force from the prop and that ‘true’ biped probably needs more processing power to walk smoothly too, so much kinematic modelling for each joint to deal with – remember it doesn’t take much force near the top of the robot to pivot it around each foot for ballance, there is a nice big lever.

      Clearly going to be a worse flier than something not carrying that mass and drag, but because it can ‘walk’ it can traverse areas a multirotor would find challenging – as a space gets tighter the turbulence from its own lift motors gets worse, and it has less margin for error, clip a rotor the whole thing might well be wrecked, but the walker mode can probably bash its prop many times, as they don’t have to be spinning near as fast and its less likely to hit anything anyway – as those feet rather anchor it…

      1. It is true that most bipeds have horrible efficiency to start with. I was thinking of a more “optimal walker”, google “passive dynamic walker” and “cornell ranger”. I think the Salto P1 have a better leg vs thruster ratio, where the fans only help with orientation when no foot is on the ground.

  1. Why does it need to have those legs? Yes, it can walk a tightrope but why would it need to as it can fly and as it seems to me, it can’t walk without those “flying parts”? Without those legs one can get better flight time and if you need to stay in one place for a bit, just land on some static legs/stands. Or on wheels and one can drive around. Interesting build and concept but I feel to see the point of this other than just a proof of concept. A robot that can walk on its own and then have the propellers to move it around in the air if needed seems far more useful… help me please. :)

    1. According to the Caltech article, “LEO’s lightweight legs take stress off of its thrusters by supporting the bulk of the weight, but because the thrusters are controlled synchronously with leg joints, LEO has uncanny balance.” and “Because of its propellers, you can poke or prod LEO with a lot of force without actually knocking the robot over.”

      In other words, the rotors give it better balance and make it lighter than other walking robots. At the same time its legs make it able to perch, stand and walk in more places and take load off of the rotors.

      So, rotors plus legs is really greater than the sum of the parts. If your use-case requires jumping up a meter or more onto uneven terrain you’d need a much heavier and stronger legged robot with complex balancing behavior or you’d need a drone with adaptable landing gear. This multi-modal robot solves that problem neatly.

        1. More energy consumption than what?

          Your pure multi-rotor if all you did is fly is going to beat one hauling legs around, but your endurance is going to be pretty tiny – 30 mins flight time is pretty good for a multi-rotor…

          A pure walking robot entirely balancing off its feet will be a much heavier build, as to really balance actively needs more complex leg joints, more powerful leg motors, all that weight adds up to needing to expend more energy to move the leg, quite possibly more energy to move it than this thing spends with its rotor assisted balance. And that isn’t counting the increased processing consumption – to actively balance on legs needs rapid calculations of the kinematics of the situation, to drive all the right joints to the right position – all a drone needs is a very simple combination of balance sensor and PID loop, which the legged machine would have to have as well to know which way its falling and adjust its legs controllably…

          Then there is the mobility this platform offers, legged robots that can actually deal with stairs are almost non-existent as creating that degree of foot travel to place a foot on the next step is actually quite hard – its not easy to make a durable, powerful, fast acting and flexible joint for every joint required to do so, and the joint designs that do work for lifting a foot a few extra inches to deal with steps don’t scale well, so you end up having to do that work again when you need a different sized platform – this doesn’t have that problem in the same way…

          Note I am not saying this is actually better for all things than more traditionally balanced legged robots – just that it does have some advantages to them.

          1. Birds do not use wings to stabilize when they walk. But we are right, a rigorous modelization is need… well I’ve not time to read research papers, I suppose they address this point

          2. In their technical paper there is a section entitled “Energy efficiency and cost of transportation of LEO” which discuss the point, design trade-offs and next challenge. Other related citation: ” . Achieving efficient walking with lightweight legs similar to LEO’s is still an open challenge in the field of bipedal robots, and it remains to be investigated in future work.”

    1. As a proof of concept, I think it pretty cool. Add arms for manipulation, camera, sensors, and you have a versatile robot. Fly over terrain to quickly get get where you are going, then be able to crawl/walk into areas otherwise that would inaccessible without ‘boots on the ground’ so to speak. All kinds of possibilities. Also I would think ‘walking’ uses less energy than flying as well. Think of robots like the ‘Transformer’ (in the movies) or beetles of the insect world. I think the concept has quite a bit of ‘potential’. You got to take baby steps before you walk….

Leave a Reply to PhillipCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.