Acrobatic XRL robot takes cliffs and valleys in its stride


May 23, 2013

UPenn's XRL robot jumping and grabbing the edge of a cliff five times its own height (Photo: UPenn)

UPenn's XRL robot jumping and grabbing the edge of a cliff five times its own height (Photo: UPenn)

Image Gallery (6 images)

Most land-dwelling animals with skeletons (exo or endo) have the ability to jump. It is of particular importance to survival, as running primarily consists of a long series of jumps. Without the ability to jump, a robot's freedom to move around is limited, something that is particularly true of smaller robots for which even relatively narrow trenches or low walls can prove too much of an obstacle. A robotics group at the University of Pennsylvania (UPenn) has taught a six-legged crawling robot to jump, giving it remarkable acrobatic capabilities.

There are a number of different approaches to get robots jumping, such as the Sandia hopper and the Sand Flea, which rely on a secondary mechanism that can be called upon when their primary means of locomotion is found lacking. Others, like the grasshopper-inspired and water strider-inspired microbots, have taken the biomimicry route, while Harvard's explosively driven hopper takes another approach again.

WIth their X-RHex Lite (XRL) robot, the UPenn robotics group devised yet another technique that gives the crawling robot the ability to not only cross gaps with a single jump, but also climb cliffs using single or sequential jumps, or leap a wide gap by taking a running start.

The XRL weighs 6.7 kg (14.8 lb), and has a body that stands about 20 cm (8 in) high and is 51 cm (20 in) long. The robot has six legs shaped into half circles, which are made of fiberglass with rubber blocks on the outside to provide grip on slippery surfaces. The springiness of the legs allows them to store energy from an earlier maneuver, a capability that greatly increases the XRL's ability to jump and run.

Each leg is powered by an independently controlled Maxon 50 watt brushless motor, providing a great deal of flexibility in how the XRL maneuvers. The frame is made of an interlocking series of flat waterjet-fabricated aluminum shapes, while a carbon fiber shell covers the frame, providing protection for the robot's machinery as well as acting to stiffen the XRL against the loads driven by the legs. The electronics include a battery compartment, a control computer, and three electronics units, each of which controls two motors.

The XRL is capable of some remarkable feats, such as is shown in the figure above, in which the XRL successfully leaps a 60 cm (24 in) gap by taking a running start. The robot can also perform a 30 cm (12 in) vertical standing jump, make a climbing double leap onto a ledge 29 cm (11.4 in) high, and vertically leap-grab onto a desk 73 cm (29 in) high.

The UPenn team presented the XRL robot at this year's IEEE International Conference on Robotics and Automation and, in the near-term, aims to find out how to use the XRL's capabilities, and in particular the energy storage in its compliant legs, to enable new forms of acrobatic motion. Such robots seem destined to reach heady heights, particularly in autonomous applications.

The video below shows the XRL in full acrobatic mode.

Source: University of Pennsylvania Kod*lab

About the Author
Brian Dodson From an early age Brian wanted to become a scientist. He did, earning a Ph.D. in physics and embarking on an R&D career which has recently broken the 40th anniversary. What he didn't expect was that along the way he would become a patent agent, a rocket scientist, a gourmet cook, a biotech entrepreneur, an opera tenor and a science writer. All articles by Brian Dodson

Boston Dynamic's Sand Flea Jumping Robot is alot more impressive IMO.


This thing should be called: Acrobot.

What amazes me is that people get paid to do this sort of work. There is probably some military potential and funding in this research.

David Clarke

@Fusionmkx I'm inclined to agree, and the method used by the Sand Flea Jumping Robot should be implementable in this robot as well, but this method has benefits the Sand Flea does not.

The main strength of teaching this kind of robot to jump and leap the way they have is that it is much faster at clearing small gaps and jumps than the Sand Flea - it can make the jump without stopping, and keep going immediately. It's also much easier on the machine's components, which in the Sand Flea must withstand two instances of very high G-forces for every jump.

Joel Detrow
Post a Comment

Login with your Gizmag account:

Related Articles
Looking for something? Search our articles