Tag Archives: DARPA Robotics Challenge

A Walk in the Woods With Boston Dynamics’ Atlas Robot

Peter Lobner

The DARPA Robotics Challenge (DRC) Finals held in June 2015 demonstrated the rather limited capabilities for state-of-the-art robots, all of which required teleoperators (remote operators) to augment limited autonomous capabilities aboard the robots. One criticism of that competition was that the original rules got watered down because of the limitations of the robot competitors. Performance of the robots could be characterized as slow and deliberate. None of the robotic competitors that fell over could get up and one was decapitated by the fall. Here’s a video compilation of robots falling during the 2015 DARPA finals:

https://www.youtube.com/watch?v=g0TaYhjpOfo

Team KAIST won the competition with their Hubo robot, which didn’t fall, but wasn’t designed to recover from a fall. Team IHMS Robotics placed second in the competition with their Running Man robot, which was based on the Boston Dynamics Atlas robot. Several other teams also based their entries on the Atlas robot. See my 2 July 2015 post on the DRC Finals.

In February 2016, Boston Dynamics posted a video of a new version of their Atlas robot, which they describe as follows:

“A new version of Atlas, designed to operate outdoors and inside buildings. It is specialized for mobile manipulation. It is electrically powered and hydraulically actuated. It uses sensors in its body and legs to balance and LIDAR (Light Imaging, Detection And Ranging) and stereo sensors in its head to avoid obstacles, assess the terrain, help with navigation and manipulate objects. This version of Atlas is about 5′ 9″ tall (about a head shorter than the DRC Atlas) and weighs 180 lbs.”

New version of BD AtlasNew version of Atlas. Source: Boston Dynamics

The autonomous balancing capabilities of this new version, especially its ability to recovery from upsets, seem significantly better than anything seen during the DRC. Atlas recovered nicely from the slip in the above photo. You can see the new version of Atlas perform in the Boston Dynamics video at the following link:

https://www.youtube.com/watch?v=rVlhMGQgDkY

Another interesting new robot from Boston Dynamics is the quadruped SpotMini, which they describe as follows:

“SpotMini is a new smaller version of the Spot robot, weighing 55 lbs. dripping wet (65 lbs. if you include its arm.) SpotMini is all electric (no hydraulics) and runs for about 90 minutes on a charge, depending on what it is doing. SpotMini is one of the quietest robots we have ever built. It has a variety of sensors, including depth cameras, a solid state gyro (IMU, inertial measuring unit) and proprioception sensors in the limbs. These sensors help with navigation and mobile manipulation. SpotMini performs some tasks autonomously, but often uses a human for high-level guidance.”

BD SpotMiniSpotMini. Source: Boston Dynamics

On 23 June 2016, Boston Dynamics posted the following short video of SpotMini in action.

https://www.youtube.com/watch?v=tf7IEVTDjng

Google acquired Boston Dynamics in late 2013. Since then, Google was reorganized, with the “parent firm”, Alphabet, being created in 2015. Shortly thereafter, Google’s research and development group, formerly Google(x), was renamed simply X, or Google X. This group includes a robotics team known as Replicant.

In March 2016, Google announced that Boston Dynamics was up for sale. One reason appears to be that the Boston Dynamics robotics work did not fit in the business model planned for Google X, which has a greater focus on relatively near-term return on investment in the form of a marketable products. You can read an interesting article on Boston Dynamics being put sale at the following link to the Bloomberg Technology website:

http://www.bloomberg.com/news/articles/2016-03-17/google-is-said-to-put-boston-dynamics-robotics-unit-up-for-sale

In late May and early June 2016, several sources (Nikkei, Tech Insider, and engadget) reported that Toyota was negotiating with Alphabet for the sale of Boston Dynamics. Also part of this sale may be Google’s Japanese robotics company, Schaft, which won the 2013 DRC Trials with its S-One humanoid robot. Schaft withdrew from the 2015 DRC Finals for the declared reason of wanting to focus on commercial products. See the article on the engadget website at the following link:

https://www.engadget.com/2016/06/01/toyota-alphabet-boston-dynamics/

It will be interesting to see how and when the sales of Boston Dynamics and Schaft are completed. If these firms do wind up being bought by Toyota, then Toyota’s Research Institute should become a very powerful center for robotic development.

NASA’s Valkyrie (R5) Humanoid Robot is Being Groomed to Support Future Space Exploration Missions

Peter Lobner

The design of National Aeronautics and Space Administration’s (NASA’s) humanoid robot R5, commonly known as Valkyrie, started in October 2012 and it was unveiled in December 2013.

NASA Valkyrie robot  Source: NASA

Valkyrie was developed by a team from NASA’s Johnson Space Center (JSC) in Houston, in partnership with the University of Texas and Texas A&M and with funding from the state of Texas to compete in the Defense Advanced Projects Research Agency’s (DARPA) Robotics Challenge (DRC).  You’ll find a technical description of Valkyrie on the IEEE Spectrum website at the following link:

http://spectrum.ieee.org/automaton/robotics/military-robots/nasa-jsc-unveils-valkyrie-drc-robot

In the 2013 DRC Trials Valkyrie was a Track A entry, but it failed to score any points, largely due to unforeseen data communications problems.  An assessment of the developmental and operational problems encountered during the 2013 DRC Trials and another assessment of Valkyrie by the Florida Institute for Human & Machine Cognition (IHMC) is reported on the IEEE Spectrum website at the following link:

http://spectrum.ieee.org/automaton/robotics/humanoids/update-nasa-valkyrie-robot

Valkyrie did not compete in the 5 – 6 June 2015 DRC Finals. Instead, NASA brought two Valkyrie robots to the DRC Finals for display and demonstration and to help promote NASA’s Space Robotics Challenge (SRC), which was announced in March 2015.

NASA describes the SRC as follows:

“The Space Robotics Challenge is currently contemplated as a dual level, two-track challenge. The Level I challenge would involve a virtual challenge competition in software simulation and the Level II demonstration challenge would involve use of software to control a robot to perform sequences of tasks. Both Levels of the challenge would have a Track A and Track B option. A competitor would pick only one track in which to compete. Track A would utilize the Robonaut 2 platform and focus on simulated in-space tasks such as spacecraft maintenance and operations in transit to Mars, while Track B would utilize the R5 platform robot to perform simulated tasks on planetary surfaces, such as precursor habitat deployment on Mars, or disaster relief in an industrial setting on Earth.”

The highest scoring teams from the Level I (simulation) challenge will be given access to NASA-provided robots to prepare for the Level II (physical) challenge.

You can download a NASA Fact Sheet on SRC at the following link:

https://www.nasa.gov/sites/default/files/atoms/files/fs_space_robotics_150908.pdf

As part of SRC, NASA awarded Valkyrie robots to two university groups that competed in the DRC Finals. The winners announced in November 2015 were:

  • A team at MIT under the leadership of Russ Tedrake. Team MIT placed 6th in the 2015 DRC Finals with an Atlas robot built by Boston Dynamics
  • A team at Northeastern University under the leadership of Taskin Padir, who formerly was Co-PI of the Worcester Polytechnic Institute (WPI) – Carnegie Mellon University (CMU) team that placed 7th in the DRC Finals with an upgraded Atlas robot known as Warner.

Each team has possession of a Valkyrie robot for two years; receives up to $250,000; and has access to onsite and virtual technical support from NASA. NASA stated that, “The robots will have walking, balancing and manipulating capabilities so that future research may focus on the development of complex behaviors that would advance autonomy for bipedal humanoid robots.” These two teams will not compete in the SRC Level I challenge, but will be eligible to compete in the Level II challenge.

An assessment of Valkyrie’s potential roles in future missions to Mars was published in 23 June 2015 on the IEEE Spectrum website. You can read this article at the following link:

http://spectrum.ieee.org/automaton/robotics/humanoids/nasa-wants-help-training-valkyrie-robot-to-go-to-mars

The types of activities a humanoid robot might perform on a Mars mission are expected to become tasks to be demonstrated by each team choosing Track B in the SRC.

In the time between the DRC Finals and the SRC Level II competitions, I’m sure we’ll see substantial improvements in humanoid robot performance.

2015 DARPA Robotics Challenge Results

Peter Lobner

Twenty-three teams competed in the 2015 DARPA Robotics Challenge, which was held June 5-6, 2015, at Fairplex in Pomona, CA.  The winner was Team KAIST from the Republic of Korea, and its robot DRC-Hubo.

drc-hubo-standing-rolling-1433820657680 Source: DARPA

An assessment of how DRC-Hubo won the 2015 DARPA Robotics Challenge appears on the IEEE Spectrum website, which you can read at the following link:

http://spectrum.ieee.org/automaton/robotics/humanoids/how-kaist-drc-hubo-won-darpa-robotics-challenge

The results of the 2015 DARPA Robotics Challenge are listed in the following table:

2015 DARPA Robotics Challenge results  Source: DARPA

You can view information on all of these teams and their robots on the DARPA website at the following link:

http://www.theroboticschallenge.org/teams

Congratulations to all teams for their efforts in advancing robotics technology and special congratulations to Team KAIST for winning the 2015 DARPA Robotics Challenge.

28 July 2016 Update:  What was the outcome of DRC?

I looked for a good explanation of the outcome of the 2015 DARPA Robotics Challenge and didn’t find much of an answer until I came across the 6 July 2015 article entitled, “The DARPA Robotics Challenge was a Bust,” posted on the Popular Science website at the following link:

http://www.popsci.com/darpa-robotics-challenge-was-bust-why-darpa-needs-try-again

The author, Erik Sofge, noted that, in 2012, DARPA originally proposed the following tasks for the DRC:

  1. Get into a standard human vehicle and drive it to a specified location.
  2. Get out of the vehicle and travel across rubble.
  3. Clear obstacles from a doorway.
  4. Open the door, and enter the building.
  5. Find a leaking pipe and close the associated valve.
  6. Reconnect a hose or cable.
  7. Climb a ladder.
  8. Grab a tool from the site, break through a concrete wall and exit.

At the 2013 DRC Trials in Miami, poor robot performance indicated the need to redefine most of these tasks. For the DRC Finals, all but one of the eight tasks had been greatly simplified. In his article, Eric Sofge provides a task-by-task breakdown of the DRC Finals, which I think you will find quite revealing.

Eric Sofge suggests a logical follow-on to DRC, which would be another competition with the following attributes:

  • Robots are required to demonstrate that they can fall, recover on their own, and demonstrate durability.
  • New tasks, including tasks closer in complexity to the originally proposed DRC tasks, should significantly raise the bar for expected robot performance.

It’s interesting to note that the NASA / JPL robot Valkyrie, which earned zero points in the 2013 DRC Trials, did not compete in the 2015 DRC Finals. However, Valkyrie was present at the DRC Finals to promote the new NASA Space Robotics Challenge (SRC). Valkyrie (aka R5) and another NASA robot, Robonaut 2, are being groomed to support space exploration. See my 6 July 2016 post for more information on the NASA SRC.

25 Teams From Around the World to Compete in DARPA’s 2015 Robotics Challenge Finals

Peter Lobner

20150222DRCFinalsLogo

The international robotics community has turned out in force for the DARPA Robotics Challenge (DRC) Finals, a competition of robots and their human supervisors to be held June 5-6, 2015, at Fairplex in Pomona, Calif., outside of Los Angeles. In the competition, human-robot teams will be tested on capabilities that could enable them to provide assistance in future natural and man-made disasters. Fourteen new teams from Germany, Hong Kong, Italy, Japan, the People’s Republic of China, South Korea, and the United States qualified to join 11 previously announced teams. In total, 25 teams will now vie for a chance to win one of three cash prizes totaling $3.5 million at the DRC Finals.

TeamROBOTISRobotSoloTeam ROBOTIS entry from Korea

You can see photos of other competitors and read more about the challenge at the following links:

http://www.darpa.mil/NewsEvents/Releases/2015/03/05.aspx

and

http://www.theroboticschallenge.org

The Cylons are coming!