ENG-466 Distributed intelligent systems: Course Project

This file aims to give an overview of the code implemented in the framework of ENG-466 course project.

Team members:

• Nicolas Marbot
• Christopher Stocker
• Xavier Suermondt
• Eliott Zemour

Structure of the directory:

Initial_Material/
    ├─ controllers/
    │  ├─ localization/
    │  │  ├─ localization_controller.c
    │  │  ├─ localization_super.c
    │  │  ├─ odometry.c
    │  │  ├─ odometry.h
    │  │  ├─ kalman.c
    │  │  ├─ kalman.h
    │  │  ├─ trajectories.c
    │  │  ├─ trajectories.h
    │  │  ├─ linear_algebra.c
    │  │  ├─ linear_algebra.h
    │  ├─ flocking/
    │  │  ├─ flocking_controller.c
    │  │  ├─ flocking_super.c
    │  ├─ formation/
    │  │  ├─ formation_controller.c
    │  │  ├─ formation_super.c
    │  ├─ crossing/
    │  │  ├─ crossing_left_controller.c
    │  │  ├─ crossing_left_super.c
    │  │  ├─ crossing_right_controller.c
    │  │  ├─ crossing_right_super.c
    ├─ worlds/
    │  ├─ localization.wbt
    │  ├─ crossing.wbt
    │  ├─ obstacle.wbt
    │  ├─ crossing-test.wbt
    │  ├─ obstacle-test.wbt
    
PSO/
    ├─ controllers/
    │  ├─ pso_obstacle_super.c
    │  ├─ pso_obstacle_controller
    │  ├─ pso.c
    │  ├─ pso.h
    ├─ worlds/
    │  ├─ obstacle-test-pso.wbt
    

Part I: Localization Techniques for Individual Robots

The code for part I can be found in Initial_Material > controllers > localization. To run this part, open the world file localization.wbt. It should have localization_super.c as controller in the supervisor node and localization_controller.c as controller for the epuck node. This should create a csv file for the odometry, gps and kalman results: localization.csv. The supervisor writes the metric for this part in the pose.csv file.

The Matlab file that can be found at Initial_Material > matlab > plotmain_x.m should reproduce the plots presented in the report. Note that the linear_algebra.c/h files are helper files for the Kalman algorithm.

Part II: Spatial Coordination Solutions

Flocking

• Obstacle: The code for the flocking in the obstacle world can be found at Initial_Material > controllers > flocking. To run this part, open the world file obstacle_test.wbt. Then select flocking_super.c as controller in the supervisor node and flocking_controller.c as controller for the epuck nodes. The supervisor writes the metric for this part in the flocking_metric.csv file.
• Crossing: The code for the flocking in the crossing world can be found at Initial_Material > controllers > crossing. To run this part, open the world file crossing_test.wbt. XAV EXPLIQUES QUOI METTRE EN SUPER ET CONTROLLER EN FONCTION DU ROBOT ID. Then select flocking_super.c as controller in the supervisor node and flocking_controller.c as controller for the epuck nodes.

Formation

• Obstacle: The code for the formation in the obstacle world can be found at Initial_Material > controllers > leader and Initial_Material > controllers > follower. To run this part, open the world file obstacle_test.wbt. Then select formation_super.c as controller in the supervisor node and leader.c as controller for the epuck0. Followers that are the other epucks have the controller follower.c. The supervisor writes the metric for this part in the formation_metric.csv file.

Part III: Particle Swarm Optimization (PSO)

The code for the PSO part can be found at PSO > controllers. It is divided in three files that communicate with each other. The pso.c file implements the main PSO loop (initialize swarm, move particles, evaluate, find best particle). The supervisor pso_obstacle_super.c repositions the particles at the initial position at each iteration, then send candidate solutions to robots (Reynolds rules weights & thresholds), and calculates fitness. The controller pso_obstacle_controller.c receives the Reynold weights from the supervisor, and runs the controller with those weights.