Group items tagged

In its latest episode, the Robots Podcast interviews the lead researcher of the Distributed Flight Array and one of my colleagues at the ETH Zurich's IDSC, Raymond Oung. The Distributed Flight Array (DFA) is an aerial modular robot. Each individual module has a single, large propellor and a set of omniwheels to move around. Since a single propellor does not allow stable flight, modules move around to connect to each other. As shown in this video of the DFA, the resulting random shape then takes flight. After a few minutes of hovering the structure breaks up and modules fall back to the ground, restarting the cycle. As most projects at the IDSC, the DFA is grounded in rigorous mathematics and design principles and combines multiple goals: It serves as a real-world testbed for research in distributed estimation and control, it abstracts many of the real-world issues of the next generation of distributed multi-agent systems, and it provides an illustration for otherwise abstract concepts like distributed sensing and control to a general public. For more information on current work, future plans and real-world applications, read on or tune in!

It should be able to hold an inert grenade with one hand, and pull the pin with the other hand without the need for human control.  The software system must enable the robot to perform the Challenge Tasks following a high-level script with no operator intervention. For example, the operator would issue a command such as “Throw Ball.” That command would in turn decompose into a sequence of lower-level tasks, such as “find ball,” “grasp ball,” “re-grasp ball, cock arm, and throw.”

Ranger, a four legged bi-pedal robot, set an unofficial record at Cornell last month for walking 23 kilometers (14.3 miles), untethered, in 10 hours and 40 minutes. Walking at an average pace of 2.1 km/h (1.3 miles per hour), Ranger circled the indoor track at Cornell’s Barton Hall 108.5 times, taking 65,185  steps before it had to stop and recharge. Ranger walks much like a human, using gravity and momentum to help swing its legs forward, though its looks like a boom box on stilts. Its swinging gait is like a human on crutches since the robot has no knees, and its two exterior legs are connected at the top and its two interior legs are connected at the bottom.

Carnegie Mellon has taught its robotic snake to climb trees, though one hopes it won’t start offering your spouse apples. “Uncle Sam” (presumably named for its red, white, and blue markings) is a snake robot built from modular pieces. The latest in a line of ‘modsnakes’ from Carnegie Mellon’s Biorobotics Lab, Uncle Sam can move in a variety of different ways including rolling, wiggling, and side-winding. It can also wrap itself around a pole and climb vertically, which comes in handy when scaling a tree. You have to watch this thing in action. There is something incredibly life-like, and eerie, about the way it scales the tree outdoors and then looks around with its camera ‘eye’. Projects like Uncle Sam show how life-mimicking machines could revolutionize robotics in the near future.

Scientists at the Johns Hopkins University Applied Physics Laboratory (APL) were awarded no less than $34.5 million by the Defense Advanced Research Projects Agency (DARPA) to continue their outstanding work in the field of prosthetic limb testing, which has seen them come up with the most advanced model yet. Their Modular Prosthetic Limb (MPL) system is just about ready to be tested on human subjects, as it has proved successful with monkeys. Basically, the prosthetic arm is controlled by the brain through micro-arrays that are implanted (gently) in the head. They record brain signals and send the commands to the computer software that controls the arm. To be honest, it will be interesting to see just how these hair-chips are attached to the brain, but the APL say clinical tests have shown the devices to be entirely harmless. The monkeys didn’t mind them too much, at least.

SLAM ( Simultaneous localization and mapping ),  PID ( Proportional integral derivative ) controller & ODOMETRY ( hodos, meaning “travel“, “journey” and metron, meaning “measure“)

The latest episode of the Robots Podcast looks at the following scenario: Imagine being able to throw a hand-full of smart matter in a tank full of liquid and then pulling out a ready-to-use wrench once the matter has assembled. This is the vision of this episode's guests Michael Tolley and Jonas Neubert from the Computational Synthesis Laboratory run by Hod Lipson at Cornell University, NY. Tolley and Neubert give an introduction into Programmable Matter and then present their research on stochastic assembly of matter in fluid, including both simulation (see video above) and real-world implementation. Read on or tune in!

Robot motion planning has traditionally been used to avoid collisions when moving a robot arm. Avoiding collisions is important, but many other desirable criteria are often ignored. For example, motions that minimize energy will let the robot extend its battery life. Smoother trajectories may cause less wear on motors and can be more aesthetically appealing. There may be even more useful criteria, like keeping a glass of water upright when moving it around. This summer, Mrinal Kalakrishnan from the Computational Learning and Motor Control Lab at USC worked on a new motion planner called STOMP, which stands for "Stochastic Trajectory Optimization for Motion Planning". This planner can plan paths for high-dimensional robotic systems that are collision-free, smooth, and can simultaneously satisfy task constraints, minimize energy consumption, or optimize other arbitrary criteria. STOMP is derived from gradient-free optimization and path integral reinforcement learning techniques (Policy Improvement with Path Integrals, Theodorou et al, 2010).

Marius Muja from University of British Columbia returned to Willow Garage this summer to continue his work object recognition. In addition to working on an object detector that can scale to a large number of objects, he has also been designing a general object recognition infrastructure. One problem that many object detectors have is that they get slower as they learn new objects. Ideally we want a robot that goes into an environment and is capable of collecting data and learning new objects by itself. In doing this, however, we don't want the robot to get progressively slower as it learns new objects. Marius worked on an object detector called Binarized Gradient Grid Pyramid (BiGGPy), which uses the gradient information from an image to match it to a set of learned object templates. The templates are organized into a template pyramid. This tree structure has low resolution templates at the root and higher resolution templates at each lower level. During detection, only a fraction of this tree must be explored. This results in big speedups and allows the detector to scale to a large number of objects.

The main focus for Bastian's work was on the feature-extraction process for 3D data. One of his contributions was a novel interest keypoint extraction method that operates on range images generated from arbitrary 3D point clouds. This method explicitly considers the borders of the objects identified by transitions from foreground to background. Bastian also developed a new feature descriptor type, called NARF (Normal Aligned Radial Features), that takes the same information into account. Based on these feature matches, Bastian then worked on a process to create a set of potential object poses and added spatial verification steps to assure these observations fit the sensor data.

EARL (that would be, Enhanced Automated Robotic Launcher) is a second generation bowling robot. Let me explain to you why a bowling robot is necessary at all: apparently, EARL is “invaluable in the many studies necessary to keep up with the ever-changing bowling ball industry… [EARL is] the future of bowling research.” Hmm. That’s worth pondering for a minute or two. Well, I won’t pretend to understand it, but that doesn’t mean I’m not impressed with the fact that EARL can throw bowling balls at 24 MPH and spin them up at 900 RPM, much faster than a human is capable of, which I guess is why people seemed to assume that the robot would win… Does it say something about the state of robotics, or something about the sport? Either way, bowling seems like a game with an extremely limited amount of random variables, and sooner or later (probably sooner) the only thing worthy of a news story will be a robot arm not bowling a perfect 300.

This is a site for all things about amateur Unmanned Aerial Vehicles (UAVs). Use the tabs and drop-down menus above to navigate the site. These are our Arduino-based open source autopilot projects: * ArduPilot, a low-cost autopilot system for planes. * ArduCopter, a fully-autonomous quadcopter system (heli autopilot coming soon). * BlimpDuino, an autonomous blimp with both infrared and ultrasonic guidance.

The Russian policeman waved at the orange van zigzagging around the empty plaza, ordering it to stop. The van didn’t, so the officer stepped closer to address the driver. But there was no driver. The van is an autonomous vehicle developed at the University of Parma’s Artificial Vision and Intelligent Systems Laboratory, known as VisLab. Crammed with computers, cameras, and sensors, the vehicle is capable of detecting cars, lanes, and obstacles -- and drive itself. The VisLab researchers, after getting tired of testing their vehicles in laboratory conditions, decided to set out on a real-world test drive: a 13,000-kilometer, three-month intercontinental journey from Parma to Shanghai. The group is now about halfway through their trip, which started in July and will end in late October, at the 2010 World Expo in China. (See real time location and live video.)

Japan’s National Institute of Advanced Industrial Science and Technology (AIST) has detailed the software used to make their robot dance (see some nice photos over at Pink Tentacle) in a recent press release.  The software, dubbed Choreonoid (Choreography and Humanoid), is similar to conventional computer animation software.  Users create key poses and the software automatically interpolates the motion between them.  What makes the software unique is that it also corrects the poses if they are mechanically unstable, such as modifying the position of the feet and waist, allowing anyone to create motions compatible with the ZMP balancing method.  This is especially important for robots like the HRP-4C, where complicated motions could easily cause it to fall over.

Bipedal humanoid robots can step over obstacles and negotiate stairs where their wheeled counterparts cannot, but this comes with the risk of falling down.  Naturally, humanoid robots will never be accepted in society if they break when they fall down.  The bigger the robot, the more likely it is that it will damage itself during a fall and be unable to get up. In 2003 the HRP-2P was the first full-scale humanoid that could fall over safely and get back up, and so far remains alone; not even Honda’s ASIMO can do this.  As soon as it detected that it was falling, the HRP-2P would bend its knees and back, which helped to reduce the ground impact.  This motion, called “UKEMI”, is quite similar to how the SONY QRIO would react when falling over to reduce the risk of damaging its components.

Cooking is an art sometimes forgotten in the robotics world, but James, the PR2 robot, and Rosie, another robot from CoTeSys (Cognition for Technical Systems) in Munich have joined forces to show that robots can be of great use in the kitchen as well. They made some pretty successful-looking pancakes and used various tools around the Assisted Kitchen to show off their skills. The main chef in the experiment was Rosie, who used her broad arms and high levels of dexterity to flip and cook the pancakes. As you can see in the video, she is a bit on the slow side, but she’s also extra careful and gets it done right. She is capable of adjusting the way she pours the batter based on the weight of the bowl, demonstrating some impressive planning and a good use of her sensors, which allow the bot to recognize how much batter she has already poured.

Events such as natural disasters, military actions, and public safety threats have led to an increased need for robust robots — especially ones that can travel across complex terrain in any dimension. The ability to scale vertical building surfaces or other structures offers unique capabilities in military applications such as urban reconnaissance, sensor deployment, and setting up urban network nodes. SRI's novel clamping technology, called compliant electroadhesion, has enabled the first application of this technology to wall-climbing robots that can help with these situations.  As the name implies, electroadhesion is an electrically controllable adhesion technology. It involves inducing electrostatic charges on a wall substrate using a power supply connected to compliant pads situated on the moving robot. SRI has demonstrated robust clamping to common building materials including glass, wood, metal, concrete, etc. with clamping pressures in the range of 0.5 to 1.5 N per square cm of clamp (0.8 to 2.3 pounds per square inch). The technology works on conductive and non-conductive substrates, smooth or rough materials, and through dust and debris. Unlike conventional adhesives or dry adhesives, the electroadhesion can be modulated or turned off for mobility or cleaning. The technology uses a very small amount of power (on the order of 20 microwatts/Newton weight held) and shows the ability to repeatably clamp to wall substrates that are heavily covered in dust or other debris.

Continuing to work on a humanoid helper robot called ARMAR, the Collaborative Research Center 588: Humanoid Robots at the University of Karlsruhe began planning ARMAR-IIIa (blue) in 2006. It has 43 degrees of freedom (torso x3, 2 arms x7, 2 hands x8, head x7) and is equipped with position, velocity, and force sensors.  The upper-body has a modular design based on the average dimensions of a person, with 14 tactile sensors per hand.  Like the previous versions, it moves on a mobile platform.  In 2008 they built a slightly upgraded version of the robot called ARMAR-IIIb (red).  Both robots use the Karlsruhe Humanoid Head, which has 2 cameras per eye (for near and far vision).  The head has a total of 7 degrees of freedom (neck x4, eyes x3), 6 microphones, and a 6D inertial sensor.

Developed by Japan’s National Agriculture and Food Research Organization and other local institutions, the robot may sound boring (when compared to humanoids, for example), but it’s actually pretty cool. The main bullet points are that it automatically detects how ripe the strawberries are (which fruit is ready for harvesting) and that it cuts the stalks without damaging the strawberries.

You may think that all of those crazy robot competitions that we like to cover are just fun and games, but there’s a serious side. Really, there is. For reals. Mindful of this, ROBO-ONE held the second annual Humanoid Helper Project last weekend, where teleoperated human-sized robots completed (or attempted to complete) three seemingly simple tasks, including pouring liquid from a plastic bottle into a cup, carrying ping-pong balls on a tray, and a 30 minute endurance race. I don’t know about you, but the last two would be a bit of a challenge for me, and they certainly were for the robots: