The goal of the predator prey design project is to give you an experience of
developing an embodied & situated intelligent agent.
In a predator-prey relationship, there is at least one agent that serves as a
“food source” (i.e., the prey) and at least one agent that serves as the
“consumer” (i.e., the predator). This is a common example used in development
of softbots & robots. It provides an opportunity to study a variety of
interesting concepts, such as:
Intelligent strategies, both singular & multi-agent
Building complex behaviors from simpler ones
Coordinating competing behaviors
Tuning perceptual strategies to respond to specific environmental stimuli
relevant for situation-specific responses
It also embodies a number of important higher level behaviors that would be of
general use to an intelligent agent, such as:
Searching for a goal
Each team will be designated as either a “prey” agent or a “predator” agent.
Each team is to design a robot and program it to meet its goals. The goal of
the predator is to capture the prey: the goal of the prey is to make find its
way to the safety of its den.
The challenge will take place in an arena of approximately 6 foot X 4 foot
defined by black tape. There will be obstacles in the arena also defined by
black tape. There will be three goal areas designated by blue tape: the den,
half-court, and the far-end of the court.
1. The robot body
Your robot body can be of any morphology you choose.
No robot can start out any bigger than 7 inches X 7 inches (no restriction on
You may use any configuration of touch, rotation, and light sensors (extra
touch & light sensors are available on request).
The IR port must be oriented parallel to the ground (it cannot be pointed
upward or downward).
2. The robot programming
Your robot must use the subsumption architecture to coordinate its behaviors.
Behaviors may be primitive such as “turn left”, or the can be complex (i.e.,
behavior assemblages). Example prey and predator code is attached.
Prey agents should be programmed to seek out the safety of their den. They
should stop when their den is found. If they receive a “tagged” message via the
IR port, they should stop “dead” in their tracks (see the example code).
Predator agents should be programmed to search out and tag their prey. If they
tag their prey, they should do a victory dance and send out a “tagged” message
via the IR port (see the example code).
Agents must not go beyond the arena border and agents must not go through
3. The competition
The competition will be two rounds. The first round will be round robin. Each
head-to-head competition will be scored. The top two prey agents and top two
predator agents will then enter the second round, which will be single
Each match will begin with the prey in front of its den, and the predator on
the opposite wall.
Each match will have a time limit of 5 minutes (we may decide to change this).
Scoring: Each predator starts out with 60 points. A prey takes points away from
the predator by reaching goals areas in the arena as follows:
|Reaching the far-end
|Finding its den
To earn the points, the prey must reach the goal areas in the order shown. A
prey reaches half-court and its den when its body is more than 50% in the area.
The following penalties will be tallied against either agent:
|Going through an obstacle
|Restarting your agent
Ties will be broken based on time.