Official Sponsor of these pages[ Home | Participants | Schedule | Rules | Archives | Search | Exhibition]
Official Sponsor of these pages
This page contains detailed information regarding the latest rules for each event.
Rules Committees (Chairs in boldface) |
|
Event 1: Find life on Mars
|
Event 2: Where's the Remote?
|
Event 3: Home Vacuum
|
Event 4: Hors d'oeuvres anyone?
|
Members of the rules committee for each event cannot compete in the same event (but may compete in others).
The chairs will be completing each rules committee and then drawing up the final rules for each event. If you, as competitors, would like to have input into the event you are competing in, please contact the chair (the boldface entries in the table above) directly for suggestions, advice, etc., or forward the request to either Jim Firby (firby@cs.uchicago.edu) or Ron Arkin (arkin@cc.gatech.edu).
For information about scheduled activities and deadlines, see our Schedule page.
For your perusal, the Co-chairs of the AAAI-97 Mobile Robot Competition are providing the latest versions the rules for the four events at the upcoming conference. They are:
1. The contemporary, Find life on Mars event
2. The equally important, Find the remote control event
3. Home vacuum (topic of a previous AAAI symposium).
Here is also a fourth event that will be held at the reception and judged by the attendees:
4. Hors d'oeuvres anyone? (thanks to Dave Miller)
(Includes update on May 8, shown in bold text.)
This task is inspired by the upcoming Pathfinder Mission to Mars, as well as the tantalizing (albeit limited) possibility of life on Mars as depicted by the recent meteorite analysis. Your mission, should you decide to accept it, is to explore a large area of Mars, looking for signs of past and current life, and return the life-forms, and only those life-forms, to the lander for further analysis. From satellite imagery, we have some clues as to where life may exist, but it is up to you to make a thorough exploration before your batteries run dead.
The robot will be placed in a large physically bounded area. The arena will be approximately 1000 square feet in area (either rectangular or hexagonal), with lexan walls delimiting the boundaries (assume the robot landed in a crater). The floor of the arena will be littered with a variety of small objects. The "past life-forms" will be stationary spheres and cubes, about the size and weight of tennis balls. Currently, we anticipate having three or four different colors for each shape. The "current life-forms" will be squiggle balls (they will all be the same color). The rocks will all be larger than the other objects -- at this time, we expect they will be real rocks.
In about the center of the arena will be a small enclosed pen with two openings (the "lander"). The pen will be a square 1-2 meters in length and width, will have lexan or cardboard walls, and the openings will be about 0.5 meters wide. The openings of the pen will have small lips (~ 5 cm) to prevent the squiggle balls from escaping. The doors will be hinged at the top. One will be colored blue and the other will be red.
The robots will start adjacent to the pen. The objective is for them to collect as many of the colored objects and squiggle balls in the given time (5-10 minutes), and deposit them in the pen, sorted by mobility (that is, the opening of the pen with the blue door is for squiggle balls, the red door is for all the different colored objects). Penalties are given for mixing mobile and non-mobile objects in the pen, and for putting any of the rocks in the pen. Penalties are also given for colliding with any of the rocks (the Mars rover is fragile!)
Before the contest begins, hand-drawn maps will be provided showing where certain colored objects and large rock-like objects *may* be located. These maps, which are not guaranteed to be totally accurate, may be used by the robots to help them decide where to explore. You can enter the map information into the robots in whatever way you want.
In addition, certain areas of the arena will be "danger zones", where travel within the zone will add penalty points. The danger zones will be marked by laying colored paper on the floor (exact color TBD, depending on the natural color of the exhibition hall floor), and also by marking the location within the map. If possible, we will choose a color that will enable robots with only b/w vision to distinguish danger zones from the exhibition hall floor.
There will be several categories of robot entries. There will be separate categories (with separate awards) for single robot and multi-robot teams (depending on the number of entries, we may further subdivide multi-robot teams into 2-3 robot teams and teams of 4 or more robots). An overall winner from the single and multi-robot entries will be determined by dividing a team's total score by the number of robots in the team.
There is also a separate category for robots without manipulators. Such entries will have to get within 6 inches of an object, indicate somehow that they have spotted the object, and indicate its classification (e.g., using speech). The robot must then pause for at least five seconds (to give humans a chance to remove the object from the arena). In addition, in the non-manipulator category robots do not actually have to return the objects to the lander.
There will be multiple trials, with each trial running for a predetermined length (5-10 minutes). The intention is to structure the event so that it would be difficult for an entry to explore the whole arena in the time allotted (for example, each trial may decrease the length of the event). Points and penalties will be given as described below.
In all trials, points will be awarded for the *first* object of a given shape/color, and then a different (typically smaller, but it may be the same) number of points will be awarded for each subsequent object of that particular shape/color. This is to encourage robots to find a diversity of "life-forms". And, obviously, no additional points will be given for picking up the same object more than once in the same trial.
No modifications to the "Martian" environment will be allowed (rock-like objects, "life-forms", danger zones, surrounding walls), although modifications to the "lander" itself will be allowed for penalty points (see below).
Several trials will be held, of increasing difficulty. The task itself will not change between trials, only the complexity of the environment (and one aspect of scoring). The parameters of the environment that may be changed are: number of different colored objects (some trials may have fewer different colors); density of rocks; density of "life-forms"; accuracy of the apriori map; placement of objects (e.g., certain objects may be surrounded on three sides by rocks and/or danger zones to make it harder to get at them). Another parameter that may be altered is the ratio between the number of points awarded for the first object of a given type, and the subsequent ones. Probably, the points will be equal for the first trial, but for other trials the first object will be worth significantly more points.
Note that we will not alter all the above parameters, and certainly not all at once. We will decide which parameters to alter based on the feedback we get, if any, over the months leading up to the competition and how well teams are doing during the competition itself.
The point/penalty values listed below are our current thoughts on how the event will be scored. The point/penalty values may be refined over the coming months (and may, in fact, change from trial to trial). We encourage feedback on whether the scoring is deemed "fair".
NOTE: For practicing, squiggle balls can be purchased from The Scientific Revolution at 415-322-1876 or at your local scientific or toy stores. We will supply all objects used in the actual competition.
This event is inspired by the need for robot assistants to perform 'fetch-it' tasks in partially known environments. Imagine a robot assistant helping a handicapped person around the home. The person might ask the robot to fetch an orange, the TV remote, a cup of coffee, and so on. While the robot may not know where all of these items are initially, over time it will learn roughly where they are kept.
The event will take place in an arena that contains tables, chairs, and shelves at varying heights. Scattered throughout the area, on the floor, the shelves, and the tables, will be 12 different objects. The robot will start the event near a human sitting in a chair (i.e.. judge) who will ask the robot to fetch three items. Once these items have been returned, the human will ask for three more items. The winner will be the robot that can find and return the most items in the allotted time.
The environment will be divided into a kitchen-like area and a living-room-like area with a partial divider between them. The lexan walls from last year's contest will be used for the perimeter and the divider. The living-room area will include a TV and a sofa with coffee table between them. The kitchen area will include a sink, cutting board, and kitchen table. Certain items, such as the sofa and sink will necessarily be fake. Either or both areas may include other distractor items, such as chairs and trash cans, but the area will be mostly free space, so as to allow easy mobility. The planned layout is shown as the "LIVING ROOM" and "KITCHEN" areas of the combined venue for the Home Vacuum event and Find the Remote Event.
All the furniture listed except the trash can will be possible supporting surfaces for objects. These items will be guaranteed to be large enough to be visible on a low resolution video camera from a distance. However, they are not guaranteed to be unoccluded from the robot's initial position. Tops of supporting surfaces will be uniformly colored and will be different from the color of any object that rests on them. Pictures will be available in advance. If multiple objects are on a supporting surface, they will be well separated.
There will be twelve objects in all. Although their precise locations will not be known in advance, most objects will be constrained to lie on one of a few possible pieces of furniture. Teams are encouraged to use this context information to intelligently search the area. The objects will be:
| Object | Possible locations | Comments |
| TV remote | TV | Black with buttons, always facing up. |
| Medicine bottle | Coffee table or kitchen table | Empty |
| Pillow | Sofa | White, large, irregular shape. |
| Video tape | TV or coffee table | A prerecorded one in a colorful box with distinctive colors |
| Coke can | Kitchen table or coffee table | Will be empty. |
| Coffee mug | Anywhere | Uniformly colored, empty. |
| Cereal bowl | Kitchen table or sink | Circular |
| Rubber chicken | Cutting board | (none) |
| Fruits and veggies | Sink or cutting board | banana cucumber green pepper |
| Ketchup bottle | Anywhere | (none) |
Pictures of all twelve objects will be available well before the competition. Detailed color images from many different angles will be supplied for six objects. All twelve objects will be available to the teams to inspect, present to their robot, or whatever, when they arrive at the competition.
The robot will be started in a randomly chosen location in the rink. This location will be its "home" location to which it must return the objects. The objects will be randomly placed, but will match the location constraints given above.
Each team will rank the objects by perceived difficulty for their system. A random ordering on the objects will chosen for each team using an algorithm biased to choose easier objects first. The robot will then be given 18 minutes in which to fetch the objects in the specified order. Teams will be awarded one point for each fetched object. Teams may also opt to take the following penalties:
Here are the final rules for the vacuuming contest. Please send comments/questions to me (bonasso@aio.jsc.nasa.gov) with copies to Erann gat@robotics.Jpl.Nasa.Gov, and Sebastian thrun+@heaven.learning.cs.cmu.edu.
Thanks,
Pete
The point of this contest is to explore the usefulness of intelligence in a task that appears to only require essentially unskilled labor -- simple household vacuuming. We believe that unlike vacuuming in the service industry (factories, warehouses, etc.) home vacuuming will require sensate intelligence to deal with the humans in an everyday environment. For the AAAI contest, vacuuming robots ought to be short on vacuum mechanisms and long on intelligence. That is to say, simple suction, storage, and disposal devices are sufficient for these tasks, but the robots will probably have to make reasoned trade-offs among subtasks in real time to achieve a high score.
1) To show the value of intelligent robotics in a rudimentary service task .
2) To let competitors measure the usefulness of their approach to robot intelligence.
3) To have fun.
We have a three room house (bedroom, family room, den) with a short hallway connecting the rooms. A smart vacuuming robot sits at a charging/disposal station in the hallway where it can see humans going in and out of the rooms (in the contest, charging a waning battery should not be required, but depositing trash will). Periodically, because of the activities of the human family living there, messes appear in the rooms, and sometimes in the hallway. The robot's task is to keep the floors vacuumed with minimum interruption of the activities of the human family living there. As such it is to vacuum on demand, when it knows a room might possibly be messy and on a periodic basis.
The rooms will be setup using materials similar to last year's competition. The figure provided is a sample layout which also includes the two rooms used exclusively by the Find the Remote event. The actual layout will be determined shortly before the competition and maps will be released to all competitors upon their arrival). The door openings to the rooms may be opened or "closed" (blocked). Each room will have a set of office furniture: desks, tables, and/or chairs of the kind used last year. We are looking into couches and TV stands. Robots are not expected to vacuum under furniture.
Periodically humans will come and go in the rooms. As they do, they may or may not leave "messes". The "messes" will consist of circular piles of paper confetti between 12" and 18" in diameter. Sometimes a human (possibly a teenager) may leave a mess in the hallway. There may be more than one mess in a room.
The vacuuming robots will be restricted to carry no more than two messes worth of trash before they must "deposit" the trash at the deposit site. Simply releasing the confetti at the deposit coordinates is sufficient to constitute a deposit.
(Judges will determine the amount of penalty, depending on how much marking is done)
(First day trials). All the rooms are to be cleaned once a week. In this phase, the robots will start at the disposal station, clean the each room by vacuuming the entire area of the room, and return to the disposal point. If messes are present (there may be from zero to two messes in each room), the robot must return to deposit the trash after every two messes. No humans will be present for this phase. There will be no messes in the hallway for this phase.
Scoring.
1) Fastest time (all three rooms must have been entered
(doorway
cleared) at least once to qualify for fastest time):
Max. score -- 20.
2) Clean rooms:
Max. score (three rooms and 2 messes/room):
13 x 10 = 30
+6 x 10 = 60
+3 x 10 = 30
-------------
120.
3) Penalties: Navigation -- 5 pt penalty for each physical item bumped or hit.
4) Bonuses:
Max. score possible for Phase 1: 160 pts
(Second Day Trials). The robot starts at the disposal station. Humans will periodically enter and leave the rooms (possibly leaving messes). It's the robot's job is to keep the rooms clean of messes. Robots must not enter a room when a human is present (during this phase, humans will not enter a room while a robot is cleaning). During a given 15 minute run, humans will be present in each room for 1 minute out of every 5, except for the family room which will be occupied 3 minutes out of every 5. Robots may clean messes in the hallway while humans are present there.
The house is laid out such that the family room is immediately visible to visitors from the front entrance. Thus, to the family, a clean family room is twice as important any other room being clean.
Scoring.
1) Messes cleaned in 15 minutes (i.e., a mess is cleaned when at least 80 per cent of the mess has been vacuumed):
Max. score:
(Max. of 2 messes per room/hallway)
6 x 25 = 125
+2 x 50 = 100
--------------
225
2) Penalties:
(Third Day Trials). The robot starts at the disposal station. A human will come into the hallway and indicate to the robot that a room needs to be cleaned. The robot will then move to the room, clean any messes there and return to base to deposit the trash. While the robot is cleaning the room, one or more humans will enter. The robot must not clean while a human is in the room, but must move to the nearest wall and wait until the human leaves.
Scoring.
1) Detecting and interpreting human indication of messy room.
Max. score -- 25.
2) Clean rooms:
Max. score
3 messes x 10 = 30
2 deposits x 10 = 20
--
50
3) Ability to stop vacuuming and retreat when human is present in the room - 25 pts.
4) Innovative human-robot communication -- 50 pts. We will reward natural communication that does not involve physically touching the robot or keyboard or mouse. E.g., gesturing, or acquiring and tracking.
Max. score possible for Phase 3: 150.
5) Penalties:
Maximum possible contest score = 535 pts.
Final as of May 2, 1997
Comments to konolige@ai.sri.com
This event will occur at the AAAI main reception where there will be heavy interaction with the attendees. Judging will be conducted by the attendees and the event organizers. The goal is to provide solid refreshments to the attendees in close quarters. Safety and self-protection are paramount. A human escort (only one allowed per team within the area) will always be nearby for safety and control of the robot (i.e., if it moves out of the designated area), but is limited in their interaction with the attendees. The robots must be fully autonomous. The escort can also replenish the hors d'oeuvres on an as-needed basis.
Preliminaries will be conducted prior to the reception to ensure that the robots are safe: both from harming the attendees as well as for their own self-protection. Teams must demonstrate the following:
a) The robots will not charge into people at high speed. Some bumping may be unavoidable, but large robot platforms moving at high speed are unacceptable. Judges will have discretion here to disquality any dangerous entrants, or to ask teams to modify their robot's movement parameters.
b) The hors d'oeuvres must be solid food and cannot be "messy". Exactly what consitutes a messy food is difficult to determine. Solid, single-bite appetizers like pretzels are encouraged. Small, easily-spilled foods like peanuts are discouraged. Please let the event organizers know beforehand what type of food you will be using.
c) Robots must be able to carry some supply of food on a stable tray. The food must stay on the tray under the mild bumps that might be expected at a cocktail party.
d) Robots must be taskable to keep within a given area. That is, the robots should have some method for an escort to turn them away from a forbidden area. This should be done *without* turning off the robot's motors and moving it by hand, i.e., there should be a method for communicating with the robot to have it move in a desired direction.
e) Robots must move around, they cannot stay still except when offering food. It is reasonable to have a (move, offer food, repeat) strategy, where each phase takes some small amount of time.
Preliminaries will be held the day before the reception, at times to be arranged with the organizers.
At the reception, a large area will be available to which the robots are confined. The area will be some subset of the reception room, approximately 50 x 50 feet. It will be marked by lines on the floor so that escorts can tell their robot when to cut back.
Attendees will be milling about taking hors d'oeuvres from off the robotic servers. All robots must be capable of carrying a standard tray. Human interaction is the key to success. The robots must move about autonomously within the reception area and can interact by speech, vision, tactile, infrared, or whatever with potential servees. Personality counts here: your job is to have the robots engage the attendees as much as possible, convincing them that they are able servers.
Expect a fair number of attendees, so your robot should be able to deal with crowds.
Robots will be expected to stay in the reception area for an extended period of time, in fact, the whole reception, if possible, which will run some 2-3 hours. If you need to change batteries, you can do this.
Scoring will be based on a combination of audience appreciation (60%) and technical judging (40%).
Audience appreciation: attendees entering the area will receive one token that they can drop in a box corresponding to their favorite server at the exit of the reception. This is analogous to tipping (little batteries are a possibility). The highest scorer will receive 60 points; the rest will receive points based on what percentage of the highest score they achieve, e.g., if the lead robot gets 50 tokens, and your robot gets 10 tokens, it will receive 12 points.
Technical merit: maximum of 40 points.
a) Not bumping into people: 10 points. Robots that are careful to avoid running into people or over their toes will be awarded 10 points. Those that continuously annoy the attendees will be awarded no points. Judges will characterize in-between behavior. No points will be awarded for robots that do not move; in fact, robots *must* move in this contest.
b) Wandering: 10 points. Robots should not stay in one place, but try to wander around. No requirement to cover the entire area, however. Robots should also not get stuck in corners.
c) Staying in area: 5 points. Robots should be taskable to stay within the given area (see above). This explicitly means that escorts should *not* have to turn off the motors and drag them back.
d) Seeking out people: 5 points. Robots that can seek out people will be awarded extra ponits. Distinguishing people from walls and poles is helpful but not absolutely necessary: partial points will be given for any reasonable behavior.
e) Refills: 5 points. Robots that can tell when they need food refills will earn an extra 5 points.
f) Manipulation: 5 points. Robots that can actively offer attendees their wares will be awarded an extra 5 points.
We are really looking for robots with maximum "cute" appeal, so the scoring is biased towards audience participation. The technical points are there so that robots stay within reasonable guidelines.