RoboCup Research Opportunities

Perception Management

Implement a tunable and configurable perceptual attention system that handles the simulated vision and sound input for a player.

Coordinated, distributed strategies

Implement a set of strategies, figure out under what conditions each should be used, and figure out how to coordinate and synchronize implementation of a strategy using multiple players communicating only by spoken words.

Smooth player control transition

A player should, for example, be able to run up to a ball and start dribbling it without a disruptive transition phase, even though each action requires a different control law. Use ideas from fuzzy control (?) to manage the transition between control laws.

Spatial representation

Using narrow, cone-shaped visual input, and directional audio input, construct a local map containing positions of the ball and nearby players, both teammates and opponents. Also locate sidelines, goals, and compute the global location on the field of "self".

Agent modeling

Model the physical attributes (speed, stamina) of a player, its teammates and opponents, and the ball.

Path planning (computational geometry or learning)

Produce and implement plans for:

The player's movement around obstacles
The path of a kick
The anticipated path of an anticipated kick,

The location of players needed to pass the ball around an obstacle. i.e. to pass the ball from A to D, players B and C should position themselves so that there is a free (curved) path A-b-c-D. (X is a defensive player):

   

                D
  
            X  X  X       c ----------C
         X           X 
              A                X 
               
                      b ---------B

Back to the Local RoboCup page

This page is maintained by Mike Hewett