Projects:2017s1-160 Cooperating Autonomous Vehicles

From Projects
Jump to: navigation, search

Project 160: Cooperating Autonomous Vehicles

Project Advisors:

Prof. Michael Liebelt

Dr. Braedan Phillips


Team members:

Mitchell Testen

Nathan Cartledge

Troy Hu

Aims

Develop an autonomous robotic system that displays cooperative behaviour among a set of homogenous robotic units.


Background

Systems that show cooperative autonomous behaviour, also known as a Swarm, provide a heightened level of efficiency, reliability and independence. A swarm system allows an independent member that can otherwise function on it’s own, to interact as part of the larger system.


Motivation

The platforms for this behaviour are cognitive computing systems[1]. For which this project has aimed to produce a workable foundation for others to continue cooperative behaviour development.


Simulation Environment

Coppelia Robotics V-REP software simulator tool allowed for high level language simulation of several independent units in a well equipped environment. This permitted us to test several proposed behaviours and design code to be adaptable to the physical hardware, minimising program translation.


Dynamic Chain Formation

The developed cooperative behaviour allows the units to form a chain with designated ‘Leader’ and ‘Follower’ units to maintain a “real-time” census of members participating in a formation.

The units participating in the formation take one of two roles: Leader - follows a predefined path or can be remotely controlled. Follower - tracks encoded Infrared signal containing the Leader’s unique IDl to seek and maintain a position in the formation.

A pair of Infrared receivers and a single Infrared transmitter provide the Unit with a vision system to control direction, while speed is dynamically changed using an Ultrasonic sensor module. The vision system is co-dependant with the Communication System.


Communication System

Units communicate via a dedicated WiFi network using UDP packets to minimize network traffic congestion. Packets contain the Sender’s ID, Destination ID and pre-defined messages to minimize packet size and latency. The packet structure is shown below:

[Sender, Destination, Message] Sender = 3 bytes, Destination = 3 bytes,message = 1 byte

These messages are routed by a central communication point called the “Brainbox”, that is responsible for: Real-time system diagnostics System state management Chain structure formation and organisation Assigns the role of each Unit


Conclusion & Future Work

In conclusion we successfully developed an autonomous cooperating behaviour program. From testing the code on the built hardware, it was able to demonstrate individual parts of the targeted behaviour, however it lacked the processing power to fully demonstrate complete system functionality.

The system can be improved in future by: Implementing the behaviour on faster computing hardware Improving self correction of system behaviour Programming additional behaviour functions