RESCUE ME A defense research challenge envisions robots, humanoid or otherwise, that can work in disaster zones.
By JOHN MARKOFF
The New York Times
In the event of another disaster at a nuclear power plant, the first responders may not be humans but robots. They may not even look humanoid.
The Pentagon’s research and development agency is to announce a competition on Tuesday to design specialized robots that can work in disaster zones while operating common tools and vehicles. And while such tasks may well inspire humanoid designs, roboticists say they may also lead to the robotic equivalent of the Minotaur — a hybrid creature that might have multiple arms and not just legs but treads. Rumors of the challenge have already set professional and amateur robot builders buzzing with speculation about possible designs and alliances. Aaron Edsinger, a founder of Meka Robotics in San Francisco, said he was speaking with fellow roboticists around the country and was considering a wide array of possible inspirations.
“Analogs to animals such as spiders, monkeys, bears, kangaroos and goats are useful inspiration when considering parts of the challenge,” he said.
In the Tuesday announcement, the Defense Advanced Research and Planning Agency, or Darpa, lists eight likely tasks the robot will need to perform — among them driving a vehicle to a simulated disaster site, moving across rubble, removing rubble from an entryway, climbing a ladder, using a tool to break through a concrete wall, finding and closing a valve on a leaking pipe, and replacing a component like a cooling pump.
Mr. Edsinger said the challenge would be not in completing any one of the tasks but rather in integrating them into a single mission. “I feel we have already have systems that can achieve each individual task in the challenge,” he said.
The idea for the competition came from the Fukushima nuclear disaster in Japan a year ago, said Gill Pratt, a program manager in Darpa’s defense sciences office. “During the first 24 hours,” he added, “there were things that should have been done but were not done because it was too dangerous for people to do them.”
The agency has not yet announced how much it intends to spend on the program or the size of the prize. It is calling the program a “robotics challenge,” which is distinguished from a series of “grand challenge” events it held in 2004, 2005 and 2007, with $1 million and $2 million prizes for a contest to design autonomous vehicles to drive in desert and urban settings.
Corporate and university teams will compete to enter the robots in contests in 2013 and 2015. The robots will not need to be completely autonomous, but rather will be “supervised” by human operators, much as ground-based pilots now fly military drones.
The competition underscores the rapid progress being made in autonomous systems in military, manufacturing and home applications. Robotics researchers have said that these advances are largely a result of the falling cost of all kinds of sensors, as well as developments in perception technologies that make it possible for robots to move in unplanned environments.
A number of ambitious humanoid robots have already been designed by industrial researchers. The Honda Asimo was unveiled in 2000 and by 2005 could operate for a full hour on batteries. Last year it demonstrated the ability to run as fast as six miles an hour.
Darpa officials said they were hoping for international participation in the robot competition. Indeed, the challenge echoes a proposal made in November by Hirochika Inoue, the father of humanoid robot development in Japan.
Despite Japan’s significant investment in robotics, he noted that the country did not have any robots capable of completely replacing humans at the time of the Fukushima disaster.
“Many people wanted to do it by robots,” he said in an e-mail, “but we had not prepared.”
In the United States, both General Motors and Boston Dynamics, a small research lab financed by the military, have developed humanoid robots. G.M.’s Robonaut 2 is now on the International Space Station, where it is being tested as an astronaut’s assistant. Boston Dynamics, which has attracted attention for a transport robot called BigDog and more recently for a four-legged running robot called Cheetah, has a humanoid robot called Atlas.
In its announcement, Darpa says it will distribute a test hardware platform with legs, torso, arms and head to assist some of the teams in their development efforts. Several robot researchers said a version of the Boston Dynamics Atlas was a likely candidate for this role, but Mr. Pratt said his agency would also provide a software simulator to allow the widest possible participation in the challenge.
“We’re opening the aperture as wide as we can,” he said.
In the third VRC task, the robot had to pick up a hose, connect it to a pipe and turn a valve
The New York Times
After Several Mergers, Seven Teams to Receive DARPA Support to Compete with an ATLAS Robot in 2013 DARPA Robotic Challenge Trials
The DARPA Robotics Challenge (DRC) was created with a clear vision: spur development of advanced robots that can assist humans in mitigating and recovering from future natural and man-made disasters. Disasters evoke powerful, physical images of destruction, yet the first event of the DRC was a software competition carried out in a virtual environment that looked like an obstacle course set in a suburban area. That setting was the first proving ground for testing software that might control successful disaster response robots, and it was the world’s first view into the DARPA Robotics Challenge Simulator, an open-source platform that could revolutionize robotics development.
Disaster response robots require multiple layers of software to explore and interact with their environments, use tools, maintain balance and communicate with human operators. In the Virtual Robotics Challenge (VRC), competing teams applied software of their own design to a simulated robot in an attempt to complete a series of tasks that are prerequisites for more complex activities.
Twenty-six teams from eight countries qualified to compete in the VRC, which ran from June 17-21, 2013. DARPA had allocated resources for the six teams that did best, but in an interesting twist, good sportsmanship and generosity will allow members of the top nine teams, listed below, to move forward:
- 1. Team IHMC, Institute for Human and Machine Cognition, Pensacola, Fla. (52 points)
- 2. WPI Robotics Engineering C Squad (WRECS), Worcester Polytechnic Institute, Worcester, Mass. (39 points)
- 3. MIT, Massachusetts Institute of Technology, Cambridge, Mass. (34 points)
- 4. Team TRACLabs, TRACLabs, Inc., Webster, Texas (30 points)
- 5. JPL / UCSB / Caltech, Jet Propulsion Laboratory, Pasadena, Calif. (29 points)
- 6. TORC, TORC / TU Darmstadt / Virginia Tech, Blacksburg, Va. (27 points)
- 7. Team K, Japan (25 points)
- 8. TROOPER, Lockheed Martin / University of Pennsylvania / Rensselaer Polytechnic Institute, Cherry Hill, N.J. (24 points)
- 9. Case Western University, Cleveland, Ohio (23 points)
The top six teams earned funding and an ATLAS robot from DARPA to compete in the DRC Trials in December 2013 (DARPA is also funding several other “Track A” teams to construct their own robot and compete in the Trials). The Trials are the second of three DRC events, and the first physical competition.
In a demonstration of good sportsmanship, Jet Propulsion Laboratory, which also has a DARPA-funded Track A effort with its own robot, decided to merge its two efforts and offer the bulk of the resources it earned in the VRC to other teams. DARPA split the freed resources between the next two teams:
- The robot associated with the JPL win and some funding now goes to TROOPER (Lockheed Martin).
- Additional funds are being allocated to a newly formed team of Team K and Case Western. That team, now known as HKU, will use an ATLAS robot generously donated to it by Hong Kong University to participate in the DRC Trials in December.
Thus, in total, seven teams with ATLAS robots and DARPA support will be going to the DRC Trials, where they will compete with other teams with their own robots.
VRC teams were evaluated based on task completion and effective operator control of the robots in five simulated runs for each of three tasks (15 total timed runs) that addressed robot perception, manipulation and locomotion. The tasks included: entering, driving and exiting a utility vehicle; walking across muddy, uneven and rubble-strewn terrain; and attaching a hose connector to a spigot, then turning a nearby valve. To simulate communications limitations in a disaster zone, the VRC imposed a round trip latency of 500 milliseconds on data transmission, and varied the total number of communications bits available in each run, from a high of 900 megabits down to 60 megabits.
To conduct the VRC, DARPA funded the Open Source Robotics Foundation to develop a cloud-based simulator that calculates and displays the physical and sensory behaviors of robots in a three-dimensional virtual space, in real time. The simulator allowed teams to send commands and receive data over the Internet to and from a simulated ATLAS robot—information very similar to what would be sent between a physical robot and its operator in the real world.
“The VRC and the DARPA Simulator allowed us to open the field for the DARPA Robotics Challenge beyond hardware to include experts in robotic software. Integrating both skill sets is vital to the long-term feasibility of robots for disaster response,” said Gill Pratt, DRC program manager. “The Virtual Robotics Challenge itself was also a great technical accomplishment, as we have now tested and provided an open-source simulation platform that has the potential to catalyze the robotics and electro-mechanical systems industries by lowering costs to create low-volume, highly complex systems.”
Additional video and images of the simulation and teams’ runs are available at: http://www.youtube.com/DARPAtv
and on . . .
In the early stages of a number of recent disasters, it was clear that there were limitations to what humans could accomplish due to dangerous situations (e.g., Fukushima). Humanitarian Assistance and Disaster Relief (HADR) is one of the primary missions of the Department of Defense, and in order to be prepared for future disasters, DARPA’s goal is to spur the development of an adaptable robot that can remove humans from dangerous areas to allow issues to be mitigated from a safe location.
Focusing on a problem of social significance brings the international robotics community and their expertise together to catalyze the industry.
Existing robot systems tend to have highly specialized and limited functionality, possess limited autonomy in perception and decision-making, demonstrate limited mobility, dexterity, strength and endurance, and are used primarily in lab settings.
The DARPA Robotic Challenge will focus on developing robots that can operate in rough terrain and austere conditions, using aids (vehicles and hand tools) commonly available in populated areas. Specifically, we want to prove that the following capabilities can be accomplished:
- Compatibility with environments engineered for humans (even if they are degraded)
- Ability to use a diverse assortment of tools engineered for humans (from screwdrivers to vehicles)
- Ability to be supervised by humans who have had little to no robotics training.
Supervised autonomy is critical, as it allows simple tasks to be performed by the robot without full-time operator intervention. This will be especially important in unreliable communications environments.
Success in the DRC would mark a significant leap forward for the field of robotics. The entire robotics industry would be strengthened by raising the bar for robotic hardware, software and sensors. Additional benefits include:
- Increasing the speed of advancements in robotics
- Growing international cooperation in the field of robotics
- Attracting new innovators to the field
DRC Simulator: To facilitate robot software development, DARPA is developing an open source simulation tool: the DRC Simulator. The Simulator will be populated with models of robots, robot components, and field environments and will be made available to organizations skilled in robotic software development. This simulator will help to expand the supplier base for ground robot systems (both hardware and software), increase capabilities, and in the future will help lower acquisition costs.
Atlas: The most successful teams (Track B and C) in the Virtual Robotics Challenge will be given a humanoid robot called Atlas. Atlas is one of many robots being developed to complete the physical tasks planned for the DARPA Robotics Challenge Trials and Finals events, and requires software expertise to program the robot to accomplish the objectives.
Based on our experience in past Grand Challenges spurring important technology advances, as well as our knowledge that the required tasks are quite difficult, we have planned a sequence of three increasingly demanding events for 2013 and 2014. We do not expect immediate success, but anticipate that teams will refine their approaches over the course of the DRC to achieve success by the DRC Finals.
VIRTUAL ROBOTICS CHALLENGE , June 2013: Twenty-six teams from eight countries qualified to compete in the VRC, which ran from June 17-21, 2013. The VRC teams were evaluated based on task completion and effective operator control of the robots in simulated runs for three tasks that addressed robot perception, manipulation and locomotion. To conduct the VRC, DARPA funded the Open Source Robotics Foundation to develop a cloud-based simulator that calculated and displayed the physical and sensory behaviors of robots in a three-dimensional virtual space, in real time. The simulator allowed teams to send commands and receive data over the Internet to and from a simulated ATLAS robot—information very similar to what would be sent between a physical robot and its operator in the real world. Members of top nine software teams will move forward towards DARPA’s Robotics Trials in December 2013. Seven of these teams will receive DARPA support to compete with an ATLAS Robot. (VRC Pess Release)
DARPA ROBOTICS CHALLENGE TRIALS, December 2013: Teams with both the Atlas and custom robotic platforms will participate in the first physical competition, the DARPA Robotics Challenge Trials. At the Trials, robots will perform an array of individual disaster response operations. High-performing teams will be provided with continued funding to support participation in the final Challenge event in December 2014.
DARPA ROBOTICS CHALLENGE FINALS, December 2014: DARPA will host the final Challenge event, an end-to-end disaster-style scenario, in which teams will compete for $2,000,000 USD. Details of the final competition will be unveiled at the DARPA Robotics Challenge Trials.
Atlas is being developed by Boston Dynamics, Inc. Based on its Petman humanoid robot platform, Atlas is modified to meet the needs of the DARPA Robotics Challenge. Specifications (subject to change):
- 7 Degrees of freedom in each arm, with addtional degrees of freedom in each hand
- 6 degrees of freedom in each leg; 3 at the hip, 1 at the knee, and 2 at the ankle
- Fully featured sensor head featuring stereo vision and laser radar
- Total mass of 150kg
More details will be announced in July 2013