2016, Autonomous Driving in a Multi-Agent System
During the summar 2016, A project focusing on the developement of Autonomous Driving was carried out at Smart Mobility Lab, KTH by Patrik Barkman, Mats Barkmam and Umar Chugthai.
This project focused on the development and testing on autonomous driving.
The main goal of this project was to establish a functional infrastructure
required for self-driving vehicles. Our experimental setup included a stand
which held a camera and IR spotlight. An IR absorbing carpet was placed
underneath the stand to minimize any stray IR signatures in the area of
interest. The robots were marked with unique IR reective patterns which
would be detected via our camera. Alongside a locomotive unit, each car
accommodated an onboard computer(Beaglebone) with a wifi adapter in order
to communicate with a remote host(a desktop computer). Besides
communication, this desktop was also responsible for handling all the data
channels, namely, image acquisition, image processing and control. The image
acquisition and processing was conducted through the use of OpenCV
libraries. The algorithm for controlling multiple cars encompassed a hierarchy,
local control and global control. The local control was responsible for accurate
trajectory following for each individual vehicle, while the global control was
responsible for generating trajectories for multiple vehicles.