Javascript must be enabled for the correct page display

Machine to Machine Perception for Safer Traffic

Slijpen, G. (2013) Machine to Machine Perception for Safer Traffic. Master's Thesis / Essay, Artificial Intelligence.

ThesisGijsSlijpen.pdf - Published Version

Download (9MB) | Preview
[img] Text
AkkoordMarcoWiering.pdf - Other
Restricted to Repository staff only

Download (41kB)


Participation in traffic requires drivers to perform multitasking. A driver needs to pay attention to different events in traffic occurring at the same time. Other events, like a phone call, or people having a conversation in the back seat, can draw the driver's attention away from what is important. This thesis proposes a framework that implements a safety warning system that uses machine to machine communication to communicate with other traffic users. The safety warning system seeks to aid traffic users with the task of detecting possible dangers during participation in traffic. The system consists of four different parts: perception of the environment, the communication protocol, the localization and tracking system, and a collision risk assessment algorithm. Perception of the environment is performed by using a laser-rangefinder. The localization and tracking system keeps track of a traffic user's own location and the locations of other moving objects with respect to the traffic user itself. A single-hop broadcasting communication protocol between traffic users enhances the range of the localization and tracking system by sharing the detected moving objects with other traffic users. By using Multiple Hypotheses Tracking (MHT), measurements of the same moving object obtained by different traffic users are associated with each other. The localization and tracking system is tested by using three robots that are equipped with several sensors including a LIght Detection And Ranging (LIDAR) sensor and an Inertial Measurement Unit (IMU). The collision risk assessment algorithm consists of a Support Vector Machine (SVM) that is trained and tested using simulator data. The simulator aims to create data as realistic as possible by using a physics engine. It is shown that a collision can be detected 2 seconds before the collision occurs.

Item Type: Thesis (Master's Thesis / Essay)
Degree programme: Artificial Intelligence
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 15 Feb 2018 07:52
Last Modified: 15 Feb 2018 07:52

Actions (login required)

View Item View Item