Q:  How should new partnerships between schools be treated (typically for teams that have an aerial and ground robotic entry)?
        
A:  Treat everything as a combined entry (one web page, one paper, etc.). The 1998 entry fee for the qualifier covers your entire team for the qualifiers (1998, 1999), so no further entry fee is required since the other school is part of your team (of course if you are a newly entering team that is comprised of a partnership between schools, the basic entry fee is still required).

You will want to give the other school some billing on T-shirts and elsewhere. If you think that they have made a real significant contribution to your team's entry, then you might want to change the team name to Your-University/xyz-University, but for the time being, all my web material will simply list Your-University as the team. Even after making up the partnership, the other school could flake out and produce nothing, so I don't want to officially recognize them until they have made an actual contribution to your team, especially since they have nothing vested (entry fee) at this point and could walk away from you when the demands on their time became too excessive, for example (in the case of newly entering teams comprised of partnerships, a combined name could be used from the onset, e.g., "the ABC-XYZ team").

With regard to the requirement for a team web page, only one is required (or desired) per team. One school might take the lead in generating the primary web site and the information from the partner could be a link to the main page. We recommend that any linked information conform to the main page's "look and feel" so it seems like one web page.
 

Q:  What are ground vehicles supposed to do and can they score points to help the team?
        
A:  This is the "International AERIAL Robotics Competition" so there definitely IS an emphasis on the aerial aspect. The ground robots have the unique ability to mediate situations (put out small fires, trail a line for rescuers to follow that will lead them on a safe path through wreckage to survivors and back out, or possibly to extricate a survivor. None of these things are easily done by an air vehicle. The view of the ground robot is extremely limited, so working together, an aerial robot and a ground robot complement each other.

Here are some other examples of how ground robots would be of use: drums have fallen out of a truck and are scattered around the ground... there is smoke around them which obscures the view by the aerial robot from above. A ground robot can get in close and read the labels to identify the contents as to whether it is toxic or not. This is useful when planning ingress paths for human rescuers. Another use would be to approach hot or wet areas where the threat of losing the air vehicle is too great. A ground vehicle can be shielded from heat and water more effectively than the air vehicle because it can carry the weight of such sheilding. Also the updraft winds produced by 40-foot flames will pose problems for approaching air vehicles, but a ground vehicle would be unaffected by this.

Finally, teams with multiple agents (air and ground) working together will be able to perform the survey mission in less time. This will allow survivors to be rescued before they "die", and that is worth more points. Teams with successfully integrated aerial and ground robotic vehicles will also recieve subjective points for that aspect of their system design during static judging.

In general, one would assume that the aerial robot should be used to quick survey and reconnaissance which is then communicated to a ground vehicle which can actually do something to improve the situation or effect a rescue at locations already identified by the aerial vehicle. In some cases, the ground robot will actually be able to get a better view of a potential target than the aerial robot.
 

Q:  Can the ground vehicle locate and/or identify objects?
        
A:  Ground vehicles can locate and identify objects should they encounter them, but reporting needs to be back through the air vehicle otherwise the ground vehicle could do all the work and the air vehicle could merely hover autonomously the entire time.
 

Q:  Is there anything that the ground vehicle can specifically NOT do?
        
A:  Fly.
 

Q:  What can ground vehicles do?
        
A:  Correctly identified things (as reported from the air vehicle) could have had their origin in data received from the ground vehicle, but nonetheless will still count for points as if the air vehicle had made the determination of its own accord.

Ground vehicles must be autonomous and subordinant to the aerial robotic component of the system in that it can be directed by the air vehicle or provide cues to direct the air vehicle, but all reporting is from the air vehicle's intelligence. What this means is that the air vehicle is not merely a relay for the ground vehicle transmissions. Rather, intelligent communication needs to be taking place wherein data from the air vehicle is updated and "more informed" based on the experiences of the ground vehicle.

Example: The aerial robot detects a potential survivor at coordinates x,y, but due to obscuration of the potential target by smoke or the inability of the air vehicle to safely approach for a better look, the ground robot is instructed to inspect those coordinates while the aerial robot proceeds to search for other targets. Upon arrival to position x,y the ground robot reports that there is a survivor, but there are drums of potentially explosive material nearby at coordinates x+4, y-1. This information is uplinked to the aerial robot which is now 100 meters away inspecting another target. The precise information from the ground robot is integrated into the map being compiled by the aerial robot and is transmitted back to the human team as a seamless report.

Confidence levels could be assigned based on how well the various sensors can identify a target. A survivor detected by the aerial robot may carry a 60 percent certainty whereas a survivor detected by the ground robot (perhaps due to its ability to move in very close), may be 99 percent. The report would therefore reflect the 99 percent detection, but the human team would not necessarily need to know the origin of the data (ground vehicle vs. air vehicle).
 

Q:  The 1999 rules indicate that some interaction with objects on the field will be possible. Have you decided what that might be?
        
A:  Interaction with objects on the ground can be removal of dead bodies, laying down of trails for rescuers, or perhaps tagging targets with beacons. Be creative!
 

 
Q:  Drum labels - what is the size of the label (Length, Width)
A:  In 1998 and 1999 the drum labels will be identical to those used in the 1997 competition, except that some drums of NON HAZARDOUS materials may have no label or have an undefined label that is unlike any of the three (biohazard/radioactive/explosive) labels of particular interest to you (not trying to fool you with non hazardous material labels that are morphs of the hazardous ones).
 

Q:  What are the characteristics of the survivors and dead bodies?
>         -size
A:  Adult human
>         -clothing
A:  Not camouflage, not bright emergency colors, no Italian leisure suites.
>         -color
A:  Normal daily colors expected to be worn in temperate climate cities.  Clothing could have a predominance of black or dark colors due to dirt and soot.

>         -aural signature (i.e. "help", "where's my leg!", etc)
A:  Sounds, if present, will be adult vocal sounds ranging from moans, to speech, to screams.  Sounds, if present in a particular survivor will emanate from the immediate location of the animatronic synthetic-- that is, they may not specifically originate from the synthetic's mouth, but perhaps from a speaker in the chest or next to the synthetic.  From a distance of 5 meters, the dislocation of the sound from the location of the synthetic's mouth will not be apparent to your sensors.  Note also, there will be other conflicting sounds in the area (perhaps flames, other animatronic synthetics, water, your own engines, etc.).

 
Q:   Fire hazard - what is maximum expected size of hazard (height, width), and are they real or simulated?  If simulated, how?
A:  Large fires are planned to be propane gas jets, small fires will be burning rubble (wood/oil/tires-- mainly for smoke production).  Small fires may be allowed to burn continuously.  Propane fires will be controlled for safety and conserve fuel when not in use.  The disaster has already happened by the time you arrive on the scene, so most fires will already be burning.  It is not our intent to blast an aerial robot out of the air with a surprise explosion of flame, but modulation of flame height can occur during a run and some fires could go out during a run.  It is recommended that your robot stand off and inspect the scene before flying into it.  You would not have a priori knowledge of specific fire locations or flame dimensions were this an actual disaster as described in the rules, but from a practical point of view, the largest flames will not approximate an oil well fire, nor will they be as small as a camp fire.

 
Q:  Water fountains - what is the highest the fountains will shoot, and will we be briefed on their location?
A:  The same things apply to simulated broken water mains as to the flames (see previous Q&A).
 

Q:  Field - what is the shape of the field (dimensions), and what is the location of the LZ with respect to the field layout?
A:  Field dimensions will be much larger than the operating area for the simulated disaster, so you should never need to worry about straying out of the arena in the 1998 and 1999 events.  The landing zone (LZ) will be adjacent to one side of the arena at a safe distance from the simulated disaster.  The shape of the field will likely be rectangular, but this will be a function of which venue we select as a host.  Again, the outer boundaries of the field should be so far from the disaster scene in the 1998 and 1999 qualifiers, that it should not be of major concern.

 
Q:  Can you have multiple landings and take-offs in your accumulative score, or is each landing and take-off considered a new attempt?  If it is considered a new attempt, what prevents teams from "remembering" the info it had collected in the previous attempt.
A:  Since the infrastructure for the Millennial Event is more complex and in many ways more static than that of previous competition events, teams will not be free to practice on the actual disaster site beforehand (though provisions will be made for teams to fly in the area for equipment checkouts prior to the qualifier).  The logistics are still being formulated, but presently only one team would be allowed close to the arena at a time so that other teams making attempts later in the day could not take notes about what is in the arena and its placement.  Once a team has flown however, it will be able to continue to observe successive teams.  To keep all of the teams of a "level playing field", it is more desirable to keep a single disaster scene configuration for all teams.  Therefore, landing a taking off again is acceptable during the hour-long attempt.  For example, you may elect to have the aerial robot return to be refueled (manual refueling is OK) before it continues a different aspect of its mission.  What you learn while doing reconnaissance in the area can be remembered.  In the event that a run must be terminated after reconnaissance information has been collected, the judges may make some changes (fire locations, drum locations, survivor locations).  You will not know what, if anything has been changed.  Survivors and drums will be portable.  Multiple fire and water outlets will be provided, but not all used for a given team.  The order in which which items are changed, will be the same for each team that fouls a run and requires changes to be made.
 

Q:  What field info will we be briefed on, and when will we get this info (sooner the better)?  What will be the form of the obstacle map (paper, digital, picture, etc.)?
A:  The obstacle/target map can be in any form (paper, digital, picture, etc.) that the judges can understand, but it must use coordinate conventions as outlined in the official rules.  The field location will be provided as soon as we select the host.  Right now several governmental facilities are being explored.  Some are on the east coast of the United States, and some are in the mid west.  You will be told the field location, boundaries that are not to be exceeded, where the landing zone will be located and its dimensions.  Other restrictions particular to the area will also be noted.
 

Q:  You have points for identifying objects, and points for locating objects - does this mean we can get points for identifying an object even if we don't give a location?  And if so, what prevents competitors from simply saying "found one" randomly?  The report to the judges of having found something and giving its location - does it need a visual with it, or does it simply need a boolean and a location (number) at the ground station?
A:  Location is sufficient.  No picture of the item is required.  Note that misidentifying an object costs you points.  A judgment will be made as to whether an object was really identified based on the coordinates that you supply.  Simply identifying the correct location of something of interest without identifying it will gain points for your team, but in order to correctly locate the item, you must have had some idea that it was of interest.  Therefore one would expect that an identification would accompany any location-- but it doesn't have to for the 1998 qualifier.
 

Q:  Are there limits to what type of sensors we can use?
A:  No, so long as they are safe.  For example, placing acoustic sensors on the ground around the arena and then setting off a bomb to measure the impedance of the ground in an attempt to find objects on the surface would NOT be an acceptable sensor.
 

Q:  Are we going to be given any probability of field population of objects?
A:  No.  You would not normally have a priori knowledge of this.
 

Q:  Is the autonomous aerial robot only allowed to send the location and indentification of an object to the judges, or could it send any data that could be processed on the ground?
A:  Yes, you can have on-ground post processing of data.
 

Q:  What do you mean by briefed obstacles ?  Will we be told of some obstacles that we must not fly over ?
A:  Briefed obstacles are things that you will be made aware of before hand.  They might be telephone poles or other things not associated with the competition, but which are in the way because of the arena that we will use.

 
Q:  What is the shape of the arena or is it not known until you can fix a site. Can we assume that it will be rectangular.
A:  You are correct.  Most likely it will be rectangular, but you should never need to encounter the boundaries except along the edge where you enter the arena.  For safety, the arena will be much larger than it needs to be to house the areas of interest.
 

Q:  What material will the arms be made of, and how big are they.
A:  Human-sized and non-metallic.
 

Q:  Sound on waving arm, what kind of sound ?
A:  No sound from arm.  Sounds may come from the victim and will be recoreded human sounds at realistic human sound levels.

 
Q:  What height do the flames reach ?
A:  You won't know, but they probably won't go above 25 feet max.  Of course heat will rise far above that.  You will probably want to stand off and assess the situation first before entering the area.  That way you could guage where not to go beforehand. You shouldn't expect to have a flame blast up from under your position as most fires will be going before you arrive at the scene (i.e., all the gas mains will have already broken during the disaster and will be ignited).
 

Q:  Are the steel drums magnetic in nature
A:  Any steel drums will be magnetic, but not all drums will necessarily be steel.

 
Q:  Is landing zone fixed or can it be anywhere on the boundary of the arena.
A:  To ease logistics, we want to have a defined LZ that everyone will use.
 

Q:  If you locate something and do not identify it do we get partial credit.
A:  It probably depends on what you are identifying.  If you define a survivor as being "something of interest" but don't say what, then the judges will likely NOT award partial credit.  If you identify something as being a drum, but can not identify its contents, that that is potentially useful information and the judges would likely award paratial credit.

 
Q:  Does black color apply to both steel and plastic drums.
A:  Yes
 
Q:  What does harmless constitute for drums
A:  Any drum that has unidentified contents (no label or unrecognizable one) could be hazardous or not.  If you do identify a hazardous material in a drum (by reading its label) then that is definitely useful information that needs to be reported back for points.  Knowing that there are a bunch of drums in a location is useful information even if you don't know what is in them because in suggesting a route to the survivor for emergency personnel to follow (AD 2000 event), you might select one that avoids the "potential" hazard of the unknown drums.
 

Q:  Does a drum being labelled mean it is hazardous?
A:  No, a drum could be labeled "chalk dust", but would be non-hazardous.
 

Q:  Does an unlabelled drum mean it is not hazardous.
A:  No, its label could have been obscurred or burnt off.
 

Q:  Will there be any trees in the competition area?
A:  No, there will be "tree-like" obstacles however (e.g., poles of various kinds). 
 

Q:  Is the current version of the competion rules on the web to be considered final? I.e., is there any intention to reveal some more details about the water fountains, the fires, the survivors, the dead bodies and the kind and volume of their sounds? Although it is much more realistic to know "nothing" about their appearances, the task remains difficult even if the basic shapes are known in advance, as is the case with the drums. Especially, I do not sleep very well imagining a sudden 5m gas flame taking down our helicopter in an instant...
A:  You are right that "knowing nothing" is more realistic.  For that reason we have been intentionally vague.  If we provide information about noise levels, sizes, shapes, etc. then people will start to design their entries for just those conditions an no more.  The rules will have more information as time progresses, but the additional information will be about the logistics of the competition and not details about the size of water fountains and fires.

Regarding sudden bursts of fire, it is not our intent to blow entries out of the sky, so even though you may not know when something is going to erupt, you might fly at a distance observing the situation for a time to map where dangerous things are rather than flying in immediately.  The intent is that that disaster has ALREADY occurred, so most fires and broken pipes will already be active by the time you come on the scene.  There will be unknown events that will occur, but these will not be intended to burn the air vehicles out of the sky unless you are flying very close to the ground.  The biggest fires and water fountains will be occurring at the beginning of the run before your vehicle even enters the arena.
 

Q:  Could there be significant elevations (piles, hills) in the competetion area?  If yes, does this imply that the telemetered positions of the objects have to include their relative altitudes (x,y,z) as well?
A:  Most sites of the type that we seek are relatively flat, but some contain buildings.  An injured person could be
up on a building.  The answer is:  YES--  x, y, AND z.
 

Q:  The rules seem to suggest a unified mission within the 60-minutes time slot. What if we store the positions dertermined in previous attempts and display them at once in a subsequent one? Is this legal, and if not, how do you ensure that teams cannot make use of stored results secretely in the navigation strategy?
A:  That is NOT legal.  The location of active flames, bodies, and other items of
interest may be changed between runs.
 

Q:  What about the dimensions and borders of the arena? Are you going to define the arena layout in the updated rules, or will we have to do a quick-setup of the border lines just before the competetion?
A:  The border lines will be done just before the competition, but they will be so far back from the actual area of the simulated disaster that your vehicle should never even come close to them.  Accidentally drifting outside a boundary will not be a problem for you.
 

Q:  Does the rule against "tethers" cover things like a microphone dangling below the robot to reduce accoustic disturbances?
A:  No, that would be OK.
 

Q:  If you are planning to clarify some of the above points in the final rule version, when could we expect them to be published?
A:  Except for those areas indicated in the rules as requiring update (such as the competition site location and exact date), any clarifications will be issued not as a change in the rules, but a separate addendum to the rules (that is what you are now reading).
 

Q:  Given the larger extents of the arena, will the ceiling of 50 feet still be enforced?  We feel that an initial climb to an altitude of 50 feet or greater might give us an easier time in surveying the field.
A:  No 50 ft height restriction, however if you get too high and the judges get scared that your machine can become a threat to people and property even if it were to become ballistic, then they would abort the run.  In the middle of a large area, you could probably get up to 75 feet without anyone becoming concerned.
 

Q:  Will we be allowed to refuel during our 60 minutes of search time?  It seems from the rules that we are allowed as many sorties as we can make before all the animatronic synthetics "die", with the combination of those flights counting towards our point total.  Are we correct in inferring this?
A:  Yes, you can refuel, but that will be counted in your 60 minutes.
 

Q:  What features are you making an effort to incorporate into your animatronic synthetics?  We are currently basing our vision system on several key features of a person: head shape, skin color, and lifelike movement.  Can we count on these features being present?  What other features can we look for?
A:  The key is arm waving motion and sound (shouts? screams of pain? moaning?).  Survivors could be partially burried under rubble, they could be wet (no IR signature), and they could be covered in dirt making them black or white regardless of their true ethnicity.  Their clothing could include hoods, hats, or helmets.

 
Q:  Could you provide some more details about the arm-waving movement? Specifically, is it one arm, or two?  Are they waving from the elbow?  The shoulder? A combination?
A:  Each one (in AD 2000) will be different.  Some will probably be less elaborate and will at a minimum wave one arm (probably from the shoulder), but in an effort to keep this unstructured, I am not going to say that they won't be waving two arms or wiggling around in some other fashion.
 

Q:  Do we have any a priori knowledge about what the animatronic synthetics can be expected to be wearing?  For instance, if it were a cold climate, they might wear hats (which could then change the shape of their heads, making it more difficult for a machine vision system to identify them as people).  Also, people in warmer climates tend to wear brighter and/or more revealing clothing (showing more skin).  I have the feeling that an animatronic synthetics half-covered with debris and wearing a ski parka, gloves, and a cowboy hat would make a very difficult target to identify based on the criteria described above.
A:  Rats!  Now you've perfectly identified our survivor... we'll have to change it.  Actually, the intent is to have them look to a human observer as a person.  No outrageous clothing or attempt to hide their identity (such as with camo clothing).  They would be dressed for the location and the climate.  Survivors will be dress as expected in August in a temperate latitude.
 

Q:  What poses can the animatronic synthetics be expected to be found?  Will they all be laying down?  Or might they be sitting up?  Standing?
A:  The rules state that none are able to get out of the area under their own power (all that could are assumed to have done so already).  So survivors will be prone or sitting up.
 

Q:  Will there be random objects in the arena such as bushes or small animals (besides the debris obscuring the animatronic synthetics)?
A:  There will be other moving objects in the arena (e.g., flags, water, fire).  Not planning on bushes or small animals unless we can get some rats to feed on the dead :)
 

Q:  Is it guaranteed that at least one type of hazard/object to be detected and located will be on the field on every run. This is so that we can concentrate on one type of identification and location.
A:  There will be more than enough targets for you to amass 1000 points in 1998 and 2000 points in 1999 when taken together with your static judging score.
 

Q:  How are the live animatronic survivors going to be moving?  Is this going to be a simple arm waving motion?  What about the sounds that the survivors are going to emit?  Are all survivors going to emit sound, or just some of them?
A:  The key is arm waving motion and sound (shouts? screams of pain? moaning?).  Survivors could be partially burried under rubble, they could be wet (no IR signature), and they could be covered in dirt making them black or white regardless of their true ethnicity.  Their clothing could include hoods, hats, or helmets.

Each animatronic synthetic (in AD 2000) will be different.

The intent is to have them look to a human observer as a person.  No outrageous clothing or attempt to hide their identity (such as with camo clothing).  They would be dressed for the location and the climate.  Survivors will be dress as expected in August in a temperate latitude.

The rules state all remaining survivors are incapacitated (all that could leave under their own power are assumed to have done so already).  So survivors will be prone or sitting up.

There will be other moving objects in the arena (e.g., flags, water, fire).

 
Q:  We would like to have a prototype of an animatronic victim prior to the competition - whether this is just before or months before.  At least, it would be helpful to know more specifications about the victims.  We feel that having this prior knowledge does not take away from the realism of the contest.  After all, when a human search and rescue team is looking for human victims, they know what humans look and sound like.
A:  I can not afford to supply every team with an animatronic synthetic, and I can't give one to a single team without giving one to all (for fairness).  In this scenario, you will be looking for survivors, but unlike a search for a lost camper, you would not expect to know what any particular person looked like or what they were wearing.  Basically you are looking for signs of life.  That could be various degrees of motion, sounds of various amplitude which could range from shouting to moaning.  At times some sounds may not be audible above background noise, so plan to key on several features:  motion, sound, form (in that order).  Of course dead bodies will only exhibit the latter, but they are still worth points in the scoring formula.
 

Q:  Is the boundary of the field going to be clearly marked?
A:  Probably not.  This is because your vehicle should never need to get close to the boundary except at the starting area (where it would be marked).  For safety, the arena is going to be much larger than the specific disaster area to be searched.
 

Q:  Are we allowed to place things in the designated landing area to assist the landing of our Aerial robot?
A:  It depends on what you have in mind, but landing aids would not be out of the question as that might be part of your system of deploying and retrieving the robot under normal conditions.  Bear in mind that the set up of any landing aids would be done during your arena time and not beforehand.
 

Q:  On point 7 the 90kg limit is mass, correct? Not weight as it says in the rules?  The Hindenburg Airship "weighed" less than 90kg?
A:  Yes, the concern is how heavy will something falling from the sky be, so in the case of an air ship, we would look at the mass of the vehicle without helium.
 

Q:  No picking up a disk next year!  The competition is getting easier :)   This part should be added back, I know it probably does not fit into the mission, but it is the "classic" problem that this Aerial Robotics Competition has defined.  Everybody that has heard about the  competition remembers it as that crazy impossible task to pick up a disk.  Adding the disk pickup to the new competition would probably  distract from the new goals of the competition, but it could be added in as an extra task, maybe as a chance to get that other half of the entery fee back?
A:  Interaction with the ground WILL be in the competition, but will not need to be demonstrated until 1999's qualifier.
 

Q:  Will the simulated dead bodies be heated so that they can be detected with IR?
A:  We've considered that, but because there will be fires around, the heat difference in a dieing body and that of the heated environment will probably be an ineffictive discriminator.  Likely, your IR sensor would detect hundreds of "bodies".  Also, be advised that the injured may not be totally exposed, but could be hiding under rubble for protection with just their upper torso exposed, so a heat signature might not be easy to correlate.  In the real world, clothing also masks the human IR signature to a great degree, especially if the clothing is wet.  My guess is that motion will be your best cue. 
 

Last updated on December 16, 1998