A team of robots works together to complete errands. They fulfill orders, sort mail, and navigate all tricky grid environments. YOUR JOB is to compute collision-free paths for the team. Get
the robots to where they’re going as efficiently as possible so they
complete as many errands as possible. Use any tool you like: ML, RL, OR,
Heuristic Search. Anything goes if you think hard enough. But don’t
think too long; the clock is always ticking!
Simplified versions of such problems appear in the research literature
under the name Multi-Agent Path Finding, Multi-Robot Path Planning, and
many others. In this competition, we bridge the gap between
academia and industry by identifying the core combinatorial challenges
that make these problems difficult to solve in practice. Participation is open to individuals and teams of individuals from any
discipline and background and from any geographic location. Participating in the competition means not only demonstrating your
skills against the best in the field but also contributing to the
establishment of new benchmarks and algorithmic baselines. You'll have
the chance to make a tangible impact on the progression of the research
field, with the possibility of your innovative solutions being adopted
by the wider community. Moreover, your involvement will help increase
attention to this critical problem and contribute to the growth of our community.
2. Financial
prizes are available for distinguished performance!
Other notable performances (2nd,
3rd place in each category) will be recognized with a certificate.
Register and submit your implementation before the submission deadline
to be eligible! To lower participation costs, we also have a limited number of training awards.
Distributed according to merit, these awards come in the form of AWS
credits (up to $1000 USD), which can be used to help with the offline
development of computational models.
3. Important dates
- Warm-up round: July 13 - August 18
- Main round: August 19 - November 30
- Results Announcement: December 20