Student Project Proposal:

Agent-based Simulation of Intelligent Highways.


Supervisors

Radu Jurca
Office: INR 240
Tel: +41-21-6936711
Email: radu.jurca@epfl.ch
Michael Schumacher
Office: INR 218
Tel: +41-21-6936679
Email: michael.schumacher@epfl.ch

Project Type

Semester/Diploma project for one or two students.

Description

Motivation

Modern highways offer a very convenient and safe means of transportation. They are usually characterized by high throughput and efficient use: i.e. the debit of cars which traverse a given highway segment is high, the average speed in most situations is quite high and on the average, the speed of cars is quite constant (this eliminates unnecessary brakes and acceleration which leads to an efficient fuel usage).

However, there are some settings in which modern highways seem to perform very poorly. Above a certain degree of occupation, two very disturbing phenomena might appear:

The intuition behind this project is that adaptive speed limitations on the highway segments preceding (and including) the one where problems might appear will drastically decrease the negative effects previously discussed. Therefore the main goal of this project is to:

  1. investigate whether speeding restrictions can increase the efficiency of highways, and

  2. to determine automatic speeding restrictions that optimize highway utilization.

The methodology of the project is to use an agent-based simulation to investigate the above hypothesis and to determine optimal speeding policies. Each vehicle on the highway will be modeled by an agent which takes decisions based on a local view.

Setting

We assume that the highway is split into segments with constant length, and that on each segment we can impose one speed limitation. Constraints between the speed limit on neighboring segments have to ensure that the vehicles do not have to break too abruptly. Each segment is under the responsibility of one agent which observes the state of the highway segment (number of cars, average speed, etc) and can decide on the speed limitation for that segment. Neighboring agents can communicate (i.e. we can assume that each agent can communicate with the preceding and the following N agents).

Each vehicle is also modeled by one agent which takes decisions based on a local view: a driver wants to get to the destination as fast as possible and guides her action depending on the traffic in her immediate vicinity. We further assume that drivers respect the speed limits (within certain bounds).

Project objectives

  1. Develop a model for the driver's behavior. You might take into consideration different types of drivers (e.g. drivers which prefer to drive faster, drivers which prefer constant speed, etc). A model of the acceleration, braking and reaction time is also needed.

  2. Model the behavior of an agent that controls a highway segment (e.g. what information to collect, how to communicate with neighbors)

  3. Define efficiency criteria for the highway (e.g. throughput, average speed, fuel consumption, etc)

  4. Investigate the conditions under which the above described phenomena appear. Identify the impact of these phenomena on highway efficiency

  5. Investigate speeding policies which increase highway efficiency

Skills

Java, English

Environment

OS independent

Radu Jurca

Last modified: May 4, 2005 11:38 AM