Amenity design based on swarm control

1. Macro modeling of swarm using vector field and continuum

For a simulation of swarm with a large numbers of individuals costs so much calculation. In this research, the swarm intent is macro-modeled by a vector field, and the congestion is evaluated quantitatively by using density of continuum model. At first, a touring route in the exibition space is set as shown in the bottom left figure. And vector field f(x) is designed so that it entraines the route.
The density of the continuum is calculated based on continuity equation.

　

where u and v are defined by;

　
　

The caliculation result of density is shown in the abobe movie. The congestion is represented by density of continuum.

2. Autonomous mobile robot

To evaluate the poposed method by experiments, mobile robots are protptyped.
This robot has the folloing capabilities.
  1. StarGazer measures the position and orientation of the robot using a marker on the ceiling.
  2. The robot has infrared range sensors to distinguish obstacles.
  3. SH-2 micro computer mounted on the robot caliculates the vector field.
  4. The robot communicates with the host computer with wireless LAN.
  5. Three omni-directional actuaters enabel omnidirectional movement.
  6. Li-Po battely is mounted.
20 robots are fablicated for swarm control.

3. Optimal distribution of evacuation guidance operators

Consider that there are two evacuation routes, however, only one route is available. To guide the evacuees to the safe route, guidance operators are distributed optimally.
In this research, evacuation route and guidance operators are modeled by vector field, and operators are optimally distributed based on the optimization of vector field. The following movies show the experimental results of evacuation guidance. We can see the effectiveness of our method.
without evacuation guidance
with guidance operators

4. Optimal distribution of exhibits for amenity design

As shown in the following figures, it may be possible to reduce congestion by changing the distribution of exhibits in an exhibition space.
In this research, the human procedure of viewing exhibits is modeled by vector field and dimensional extension of dynamics is employed to represent the viewing time. The following movie shows the optimization of exhibits distribution. According to the movement of the exhibits, the number of individuals decreases in the space.


e-mail : @mep.titech.ac.jp