Home

Competitions

Arenas

Rules

Qualification

2008 Events

2007 Events

2006 Events

2005 Events

2004 Events

2003 Events

2002 Events

2001 Events

2000 Events

FAQs

Publications

Contact

Urban Search and Rescue Robot Competitions FAQs

General Administration

1. Is it possible to compete in both of the American Open and the International competitions or can we only be accepted to one of the competitions?

Answer - You may certainly compete in both regional open competitions, such as the US Open, and the World Championship competitions. In fact, if your team places in the top three teams in any regional open you are automatically qualified for the World Championships. The competitions are handled completely separately, so you'll need to submit paperwork for both. But your TDP can be the same for both competitions.

**********************************************************************
2. What does a team need to do once it has sent in its Pre-Registration form?

Answer - The next step is to download, complete and return the Team Description Template, to which is due March 1st.

**********************************************************************
3. May teams request information (nationalities, emails, robot information) about the other teams that have pre-registered?

Answer - Each team that has pre-registered for the competition will have their information (Country, Team Name, Organization/University, Team Leader, and Email) posted on the website. Team collaboration (two or more teams working together as one team) is strongly encouraged for reasons that include producing a stronger team for competition and integrating various technologies to potentially advance the state-of-the-art.

**********************************************************************
4. May teams find out earlier than the posted date if they have been selected for the competition so they can make the necessary travel arrangements (hotel, airfare,…) sooner?

Answer - No, unfortunately the technical committee will not release the final teams until the posted date because they need to have enough time review all of the Team Description Papers. It is at the discretion of each team leader as to whether or not they want to book hotel and air accommodations in advance of this knowledge.

**********************************************************************
5. Upon submitting my team description paper, will it be published or distributed to the other teams and/or the public?

Answer - Only the teams that eventually qualify and compete will have their TDP published/distributed. TDPs are not made public to anyone (except for the league chairs and technical committee) until after the competition.

**********************************************************************
6. When writing the cost section of the team description paper, are you looking for projected manufacturing cost or actual production cost?

Answer - The goal of the cost section of the TDP is to convey to next year's teams the relative costs of your approach.. This will be very important once it's been proven effective (!) and there are a few approaches to choose from. The cost estimates probably should not state, for example, the *eventual* cost of such devices, for example once the auto industry starts making them for their cars. But that is certainly worth noting in your approach -- that you see the eventuality of such reductions in cost. Next year's teams need to know how to replicate what you've done. So even for donated parts, the intent is the same. Consider somebody after the fact trying to replicate your approach and provide enough information about your system to help them find key components, purchase them, implement them, and maybe even improve upon them (but you don't need to give the entire farm away).

**********************************************************************
7. Are there any restrictions/page limitations on the Team Description Papers?

Answer - There are no restrictions or page limitations on the Team Description Paper because of the diversity of information that may be included based upon a team's current status with their platform hardware, software and approach to the urban search and rescue situation. Teams should present their current/projected approach of their system (as outlined in the Team Description Paper) as clearly and concisely as possible. Pictures and figures are strongly encouraged.

**********************************************************************
8. Is the call for symposium papers completely different from the Team Description Paper?

Answer - The Symposium papers, which are completely different than the TDPs, will be peer reviewed for inclusion in the Symposium edition of the Springer Lecture Notes on Computer Science series. They will be included either as full papers (16 page limit), or shorter poster session papers (8 page limit). These submissions must adhere to the strict Spinger rules regarding page limits (including appendices) and formatting. Detailed instructions about formatting your manuscript according to the Springer LNCS series (including style files), can be found at: http://www.springer.de/comp/lncs/authors.html. The upcoming deadline for Symposium papers is February 10, 2005. They must be electronically submitted to: http://robocupsymposium.confmaster.net/ Watch the Symposium website for details: http://www.robocup2005.org/symposium/call_for_papers.aspx

**********************************************************************
9. How many operators do you recommend be used for each mission?

Answer - Fewer is better and zero is ideal! That is, fully autonomous implementations that find even a few victims and report back accurately would likely win the competition. Take a close look at the performance metric and see that the operators are *squared* in the denominator. The goal, of course, is to simply improve upon the 1:1 ratio of operators to robots either through bounded autonomous capabilities or much better operator interfaces that allow productive control of two (or more) robots simultaneously

[Return to Top]

General Robot Information

1. Are there any restrictions on the size of robots (height, width, length, and/or weight)?

Answer - There are no official restrictions on the size of a robot. However, we are trying to project a trend toward smaller/lighter physical implementations that can be carried by a single operator. Also, there are certainly size constraints on the entire arena setup. Therefore, we have adopted the policy that major hallways will be no more than 1.2 m wide, to accommodate a 1m wide robot. But that does not mean that a 1m wide robot will be able to access all parts of all arenas. There are certainly narrow access areas, problematic rubble which can constrict hallways, and enticing passages such as ducts which can save valuable time if able to negotiate. In regards to weight restrictions, the elevated floor, ramp, and stairs can accommodate humans, so robot weight has never been a factor.

**********************************************************************
2. Is it possible for a part (or all) of the robot to fly?

Answer - Flying robots are not ruled out. In fact, we've had a couple of blimps attempt the arenas (displaying limited success). Keep in mind that there are certain rules that apply directly to flying robots. For example, there are doorways, cross-bars, and netting that restrict the available airspace at certain points. Flying robots (and ground robots) must pass UNDER these elevated obstacles to enter the next room. In many places, once through the "doorway" the flying robot may hover at a much higher altitude inside the room. The key is that all robots, including flying robots, must demonstrate the ability to effectively maneuver in a 3-D maze of obstacles. Also, keep in mind that although the application domain for these arenas is indoor mobility/navigation/mapping, that does not mean there will be no air currents. Keep in mind that collapsed buildings with broken windows can display quite windy conditions. And we've seen that ordinary climate control systems can be quite problematic for blimps.

**********************************************************************
3. Are there any restrictions to using commercially available robots as opposed to "home-made" robots?

Answer - There are no restrictions to using commercially available robots except that they must not be wider than 1.2m so that they can enter the arenas (see Question 4). Previous competitions have seen Nomad-type robots and ATRV, Jr.'s enter and negotiate the arenas (with varying degrees of success).

**********************************************************************

4. Where are the robot start points? How will my robot be faced?

Answer - Start points will be in the Yellow arena with all robots facing the same direction ("north" on your map). The start points could be at a perimeter wall, or in an interior space, but will likely have a wall nearby to latch onto for wall following behaviors. It may even start facing a wall. The actual deployment inspiration is a vertical entry through a breech on the roof/ceiling onto a lower floor (the arenas) which needs to be searched, so the robot can be placed essentially anywhere within the Yellow arena.

[Return to Top]

Arena Information

1. The overall sizes of the arenas and their distance from the operator station is important for choosing the appropriate communications devices. Will teams be provided with additional dimensional information about the arenas?

Answer - See the rules document for three different possible arena configurations (depending on the venue space available). The worst case, considering the maximum distance from one end to the other, is about 30 meters. That's if we put the operator station at one end and you make it into the far end of the red arena.

**********************************************************************
2. What are the minimum and maximum heights and widths of the stairs that may be encountered in the red and orange arenas?

Answer - The stairs' treads are typically about 750 -1200mm wide, 200-300mm deep, and about 200mm high per step. You can expect anywhere from 5-8 steps total. Keep in mind, if you're designing from scratch, that steps may be missing or there may be some rubble on the stairs. Both make it much more difficult. We don't typically do that in competitive rounds or if no robot is capable skipping a step.

[Return to Top]

Victim/Target Information

1. May teams put transmitters (for victim identification and/or robot position purposes) in the arenas prior to or during a mission?

Answer - Putting transmitters in the environment via robot is a good thing... as long as they are NOT 2.4GHz so as not to conflict with communications in the other leagues. Putting them in the environment any other way, or before each mission, is not allowed. No teams have access to the arenas prior to their mission start.

**********************************************************************
2. How many victims will be placed in each arena and how specifically will they be placed?

Answer - The victims are placed with 4 in each colored arena. They are in the boxes shown in the presentation, some wide open (autonomous and red stepfields), some through 6 inch holes. The victims are moved to different locations between rounds to ensure that during the entire competition they end up in every location, at every elevation (up to three boxes high). Each mission has two low and two high victims in the lowest two levels (the third level high victim is always out there, and counts as a high victim).

The reason for the new format was to remove any incentive/possibility of cheating by knowing in advance the location of a victim. Teams are allowed to go through and see where the victims are, plan their strategy, check for motion and heat. If you're in for multiple missions, you will see every victim placement over time and will produce an average victims found sort of metric.

**********************************************************************
3. May teams provide extra information about the victims, their condition and the arenas? Would this additional information qualify teams for bonus points?

Answer - All pertinent information about the victim is necessary, and appreciated, but already factored into the performance metric (Victim Situation, Victim State, and Victim Identification Tag. Again, pictures of the victim are welcome, but are not specifically required within the scoring metric. The more detailed information you supply, the more confident the Incident Commander will be in your victim identification and your map, and the better you will score on Map Quality and Map Accuracy.

**********************************************************************
4. Is it important to detect whether a victim's body-heat is distributed uniformly or non-uniformly over their body?

Answer - The "body heat" is accomplished by using heating blankets and pads distributed under the clothing of the victims. So it is by no means a point source, yet it does not cover every square cm of surface area on the body. For example, the mannequin head is not heated. What is important to detect is the temperature of each victim (approximately 37 degrees Celsius +/- 2 degrees).

**********************************************************************
5. Will the amount of CO2 given off by the victims be considerable or will it be as little as the realistic amount of CO2 given off by a person breathing?

Answer - The amount of CO2 you can expect is similar to human breathing. We may also put the simulated victim in a situation which allows some buildup of gas (entombed), and that generally requires a probe or other reaching device to investigate. The amount of CO2, in parts-per-million (ppm), that must be detected to receive credit is still being discussed.

**********************************************************************
6. How loud will the sounds be that are coming from the victims since there will be background noise from the spectators and other competitions?

Answer - We do not, as of now, have a specific decibel level for you to identify. The sounds are tape-recorded moans (relatively soft, but audible), screams (clearly audible), locator beacons from downed rescue personnel (ear piercingly audible), and tapping from another room or location like a basement.

**********************************************************************
7. What are the acceptable sensor ranges for detecting body heat, sound, and CO2 emission?

Answer -

Body Heat (heating pads and blankets)

Sound (audio tapes of voices, shuffling, tapping,...)

CO2 Emission (compressed tanks provide flow)

**********************************************************************

8. What size Tumbling E eye charts? Download online? Are we just supposed to show visual confirmation of these charts, report orientation, or what? For the hazmat labels are we just supposed to report what label we see, or again do we need to specify orientation and other details? Are these standard shipping box sized hazmat labels or is there a source for acquiring these?

Answer - The tumbling E's can be downloaded in pdf. They are just a few lines from a Snellen eye chart. The Hazmat labels are of a similar size, maybe 5 inch diamonds. Examples of hazmat labels can be found here. To identify either, your operator will either write the smallest line seen of Es, or circle the appropriate hazmat label icon on the victim form (sample form).

**********************************************************************

9. Why and how are victim boxes used?

Answer - The victim boxes encourage the robot to attain a particular position within the arena in order to use their sensors for victim or object identification. Sometimes manipulators are required to precisely position sensors, which can be difficult in the various terrains within the arenas and under the elevated floors. The victim box orientations are meant to differentiate the "situation" of a victim, and convey essential vocabulary that responders use. The victim box stacks closely resemble the emerging ASTM Directed Perception test methods for response robots, and contain the following targets to identify:

Generally, the boxes simply enforce a particular position of the robot for best viewing. Typically, the harder sections of the arenas have the easier victims to identify. For example, Red arena victims and Yellow arena victims that must be found autonomously have open Trapped or Void victim boxes rather than Entombed boxes which require precise sensor placements:

Victim boxes are also placed in one of three different elevations for added complexity:

In the 2007 championship, the Yellow and Red arenas had four Void victim boxes each, all open to the side placed at level 1 and level 2 elevations. The Orange arena had four Entombed victims with viewing holes, which required both reach and accuracy to see into, and one was placed placed in the highest level 3 elevation while the other three interchanged between level 1 and level 2 in any given location.

For 2008, the Yellow and Red arenas will have one Trapped victim open on top and/or one Entombed victim with a view hole. All will remain at the level 1 and level 2 elevations. The Orange arena will continue to have all types of victim boxes located in hard to reach areas like up the stairs/ramp and under the elevated floors. There will likely be at least one placed at the Level 3 elevation at all times.

**********************************************************************

10. Do teams need to fully identify and interpret victim tags, either eye charts or hazmat labels, associated with a particular victim?

Answer - Teams need to "read" whichever tag is there. It will be prominently positions, but might require a particular viewing angle to see. Once the tag is in the field of view of your camera, the operator needs to read the smallest eye chart line, or recognize the hazmat label enough to circle the appropriate label on the victim data sheet. There are five or so hazmat labels to pick from.

Even the autonomous robots must stop in front of the victim and wait for the operator and judge to read the label and fill in the sheet. But it is no more rigorous than that. You'll get full credit if the label is in the field of view and your camera has sufficient resolution to correctly interpret the tag. You may also need to deploy enough light to see the tag. Variable illumination is a hard requirement for all search and rescue robots so you don't wash out the scene.

[Return to Top]

Scoring Information

1. How does the robot need to be properly located for acceptable scoring? Physically touching some part of the confined space, full robot footprint within the space, or simply an arm that can extend into the space with or without a part of the ground transportation mechanism of the robot within the space?

Answer - In general, for all victims, the robot body needs to attain the pallet in front of the victim—entirely located on that pallet. Sometimes under the elevated floor, however, it can be hard to extend an arm that stows nicely while being underneath, and on that closest pallet also. One thing that will not be allowed is the reaching under the elevated floors from outside the confined space. But once the entire robot is working under the floors, the manipulator can be deployed in front of the robot which might not allow the entire mobility base to be completely on the closest pallet to the victim.

**********************************************************************

2. Are teleoperative robots allowed to travel through the yellow anera but just not score victims?

Answer - ALL ROBOTS MUST TRAVERSE THE YELLOW ARENA, if only because the start point is in the yellow arena, but mostly to map it even if they aren't allowed to score victims.

**********************************************************************

3. Do the victims need to be located autonomously as well as autonomous navigation?Are we still using the operator Accept/Deny system if the robot flags what it thinks is a victim and then waits for a response from the operator before continuing? Same time penalty system if it flags a false victim?

Answer - Autonomous robots must stop on the pallet in front of the victim with all sensors signatures lit up in the display... pause an ask for the operator/judge. That process will take 30seconds if they are right or wrong (penalty).

[Return to Top]

GeoTiff Information

1. What are GeoTIFF map formats? How are these maps compared to the ground truth for the arenas?

Answer - The GeoTiff is a key new requirement, specifically so we can compare to the arena ground truth setup. Maps are compared with ground truth in the attached video. The format is being driven mainly by the USARsim virtual league. For more information, go to: www.robocuprescue.org/wiki/images/GeoTiff.pdf

2. What are teams supposed to hand in at the end of a mission?

Answer - Teams should hand in a GeoTiff file on a memory stick at the end of a mission. This file will be used by the judge to compare against a GeoTiff file of the ground truth arena configuration. Administrators will not be handing out that ground truth GeoTiff file to teams. But the arena is there for all to see.

3. What are the key components of the GeoTiff file that teams should follow, since we are purposely lagging behind the Virtual Robot League in this regard to allow them to work through the issues.

Answer - Teams should adhere to the color scheme developed by the Virtual Robot League (shown below), but they DO NOT have to add the vector data that the Virtual League is working on this year:

4. Do teams still need to print a hard copy of their maps after the mission?

Answer - Teams may still print their maps, but to score the maximum points for MAP QUALITY (10 points per victim) and MAP ACCURACY (10 points per victim), teams need to hand in a GeoTiff file on a memory stick. The administrators will likely print the map and post it in any case. Any other type of map, including a printed map based on sensor data suffer in their MAP ACCURACY score since we'll not be able to quantitatively compare the map to ground truth (which is a major goal of this new exercise!).

5. Will hand drawn maps still be accepted?

Answer - Hand drawn maps will be accepted, but it won't be worth the time/effort to generate such maps since they will not score points for MAP QUALITY and MAP ACCURACY. We're not saying that topological or hand drawn maps don't have a use, but this league is about pushing the state of the art in technology for emergency responders, so such maps simply don't score very well. So don't bother drawing such maps, simply use the time to go find more victims, and show off your Best-In-Class Mobility implementation!

[Return to Top]

Network

1. Why is IEEE 802.11A (5GHz) being recommended for wireless communications instead of 802.11B (2.4GHz) or 802.11G?

Answer - There won't be any available channels in the 2.4GHz band since the other RoboCup leagues use 802.11B and 802.11G for their wireless communications.

There are also several advantages to using 802.11A as compared to 802.11B or 802.11G…

a). The 802.11A at 5 GHz would appear to have a shorter range (as would be intuitive from the frequency), but it can transmit data at a higher bandwidth (54 Mbps using 802.11A vs. 11Mbps using 802.11B). Although 802.11B appears to have great range (100m), bandwidth at that range drops down precipitously (maybe to 1Mbps). An 802.11A network has been setup at the NIST Arenas and tested within its cluttered environment to yield promising results (transmission rate between 48-54Mbps, 40m away from the access point).

b). The 802.11A uses a fundamentally different approach to slicing up the frequency range that may actually work BETTER in complex environments. That is, maintain similar ranges while supporting full bandwidth. The 802.11A was also tested, at the NIST Arenas facility, through a cinder block wall 10m from away the access point to yield a transmission rate between 24-36Mbps.

[Return to Top]

Other

1. What if I have a question that is not addressed above?

Answer - Send your question to rescue.robot.league@nist.gov and it will be answered.

isd-webmaster@cme.nist.gov
Last updated: February 29, 2008