IJCAI 2009
Robotics Exhibition
   and Workshop

Back to the Student Robotics Challenge main page

Guidelines for the challenge

Each team should plan to participate in three activites:

• The free-form exhibition and structured robotics workshop that make up the IJCAI 2009 robot venue.
• A Round 1 demonstration of their robot(s) capabilities.
• A Round 2 challenge of their robot(s) capabilities.

The following sections provide detail on each of these three activities.

Exhibition and workshop

The IJCAI 2009 robotics venue includes a two-and-a-half day exhibition open to conference attendees. It also includes a full-day workshop at which highlights of the exhibits are presented. Teams should plan to exhibit and discuss their systems and strategies during the exhibit. A short presentation at the workshop will summarize the set of entries in the Student Robotics Challenge (URC), as well.

It is customary to use the exhibition time to fine-tune and test robot platforms, sensors, and algorithms in anticipation of the two formal challenge rounds. In addition, there are conference presentations to attend and other exhibits to peruse... it's both intense and a lot of fun.

Round 1 -- "Showing Off"

In order to serve as motivation for a broad set of coursework or research investigations, the first round of the Student Research Challenge is largely of each team's own design. Here is the basic task.

The Task

• Robots will be provided a well-defined "starting spot" for exploration. If they are small, robots may be run on a tabletop. If they are large, they will run in free space within the conference venue itself.
• The team members, if they wish, may place up to 15 landmarks/objects of their own choosing and design throughout the environment around the starting spot. One landmark should designate the starting spot itself. Alternatively, teams may use landmarks "natural" to a typical indoor conference venue. This flexibility is meant to accommodate many different sensor suites for robot platforms.
• There will be static obstacles whose position will be decided beforehand through consultation between the judges and the team members. If the system requires a map of the environment, that will be possible.
• At a team's starting time, they should start their robot(s) exploring. A run will last 8 minutes.
• The goals of the run are threefold:
  • to explore as large an area and find as many landmarks as possible
  • [optional] to transport as many landmarks (or other manipulable "tokens") as possible to the starting point. The tokens may be pushed or manipulated. Also, they may be simply gathered, rather than carried back to the start.
  • [optional] to create a quantifiable representation of the environment that was explored, e.g., the landmarks' original locations. This should be provided after the end of the run.

Judging

The judging of this portion of the challenge will be based on the following five criteria, weighted equally:

• the team members' explanation of their system's implementation and its behavior while it is running
• the area, measured in robot diameters, explored and the number of landmarks found
• the robustness and generality of the system: handling obstacles and choosing landmarks
• [if manipulation is attempted] the number of tokens gathered or returned to the start
• [if a map is built] the accuracy of the environmental representation created

Other guidelines and notes for Round 1

• Time permitting, teams will have the chance to make multiple official runs of Round 1. All teams will have the time and space to make unofficial, practice runs, as well.
• Robot systems should be fully autonomous (shared-autonomy systems might consider the other IJCAI robotics exhibition tracks!) However, they need not be untethered. Wireless or wired interaction with computers is welcome; offboard sensors are welcome, as well.
• Teams may assume that their landmarks will not move during a single run, except as moved by the robot(s) themselves. However, teams should allow for at least a small number of unexpected, dynamic obstacles, e.g., people walking by.
• If robots can demonstrate additional robustness and/or generality (being "kidnapped" to an unknown nearby location, speaking the identity of each landmark, interacting with people, additional manipulation capabilities, etc.) -- DO SO. This is your chance to create conditions that will show off all of the computational intelligence that your system has to offer: Show off your system!.
• All of the judging criteria will be based fundamentally on the computational intelligence exhibited by your system. It is not in the spirit of this event to try to prohibit all of the ways in which the spirit of this paramount guideline might be avoided. Rather, we want to allow maximal flexibility to pursue - and exhibit - embodied computational intelligence!

Round 2 -- "The Unknown"

In the second round of the challenge, teams will perform precisely the same tasks as in Round 1. However, in this case, the environment will be more challenging, beccause the judges will determine the positioning of the landmarks and obstacles. This will be done in consultation with the teams, but it will also be done in such a way to push the systems to the limit of their capabilities - and perhaps beyond it. That is, teams should not expect to complete Round 2 perfectly; rather systems will be recognized by how well they do under increasingly difficult circumstances and how well they recognize conditions they can not handle.

Judging

The judging of this portion of the challenge will be based on the same criteria as for Round 1. The final comparison of the systems will include the results of both Round 1 and Round 2, with Round 2 being given twice the weight of Round 1.

Final recognition and awards

The final list of recognitions given out to teams will depend on the tasks those teams set out to demonstrate. However, we anticipate awarding at least

• Overall first-place and second-place for system excellence
• Technical award for best object manipulation
• Technical award for best spatial reasoning
• Award for best system exhibition and presentation

There may be other judges' recognitions too. Again, the fundamental criterion upon which the five judging elements build is the extent to which teams have created and demonstrated embodied computational intelligence in their entries.

Physical prizes may also be awarded - more on this as we approach the event itself.

It is also worth reiterating that we are eager to work with teams to make this a motivating, challenging, and useful event. If there are facets of the challenge that will help it dovetail with a research project, a class project, or another part of your curriculum - please let us know!

Contact the organizers

Please contact the organizers:

Deb Burhans (Canisius College) - burhansd@canisius.edu
Zach Dodds (Harvey Mudd College) - dodds@cs.hmc.edu