US mini logoHome | A-Z Index | People | Reference | Contact us
University of Sussex
About | People | Projects | Doctoral Programme | Seminar Series | Resources

POETIC: the POrtable Extendable Traffic Information Collator

POETIC (the POrtable Extendable Traffic Information Collator) is a research prototype software system, based on natural language processing and expert system technology, which accepts 'live' police reports about traffic incidents, recognises information of relevance to other motorists, formulates suitable advisory messages, and coordinates message delivery to motorists via media such as paging, cellular radio, and the Radio Data System. The project aims to demonstrate the realtime broadcasting of traffic information using live data.

The input to the POETIC system is verbatim police reports as entered into the police logging computer, and may contain abbreviations, jargon and deviant grammatical forms.

08:07:58: a24 at ewell lorry blocking road

The output from the system is the message transmitted to motorists. Messages are coded appropriately and communicated to motorists via the Vodapage paging network, RDS-TMC, AA Roadwatch, local radio, or other dissemination channels.

There is a blockage on the A24 at Ewell. The road is partially blocked. Drivers are advised to reduce speed.

The System

The POETIC system consists of six interacting modules:

The Message Analyser is a bidirectional chart parser with a domain specific lexicon and specialised grammar rules. It returns a semantic representation of each input message.

Clause1: at(e1,e3), road(e3), called(e3,a24), at(e1,e2), town(e2), called(e2,ewell), grid_ref(e2,{5221,1627}), eventtype(e1,unknown) Clause2: eventtype(e4,blockage(e5,e6)), time(e4,pres_fut), hgv(e5), road(e6)

The Incident Recogniser uses inference procedures to build and update a model of the incident. Multiple references to the same entity are resolved (as with ``road'' and ``eventtype'' in the example) and default assumptions are added (e.g. the ``degree'' clause and ``duration'' in the example).

eventtype(e1, at(e1,e2) blockage(e5,e3)) town(e2) time(e1,pres_fut) called(e2,ewell) hgv(e5) grid_ref(e2,{5221,1627} at(e1,e3) report_time('08:07:58') road(e3) duration(e1,60mins) called(e3,a24) degree(partial)

The Tactical Inferencer is a rule-based expert system which decides, using the model built by the Incident Recogniser, whether a broadcast is needed, what information should be broadcast, and to what area and for how long the broadcast should be directed.

IF incident_type = ``blockage'' AND road_type = ``M'' OR ``A'' OR ``B'' AND estimated_duration > 30 minutes THEN} generate broadcast with broadcast_desc = incident_type broadcast_target = 10 mile radius broadcast_duration = estimated_duration

The Broadcaster maps the abstract broadcast area requirement specified by the Tactical Inferencer onto actual transmission devices and formulates the broadcast message in the fashion appropriate for the selected output media (e.g natural language for AA Roadwatch, bit-string coding for RDS-TMC). It manages the frequency and timing of the broadcasts.

Description: there is a blockage Location Road Name: on the a24 Location Place Name: at ewell Degree: the road is partially blocked Advice: drivers are advised to reduce speed Transmission Device: transmitter 1 Transmission Frequency: 5 minutes

All modules of the system communicate via the Database Server. It ensures that a consistent view of incident data is maintained throughout the system and allows for the system to be distributed across a network of machines.

The System Monitor provides a user interface which displays icons representing current incidents on a map and allows for the manual entry and editing of incident records. It also permits the system operator to manually override the automatic formulation, scheduling, and targeting of broadcasts.

The Collaborators

The POETIC project involves a number of collaborators, some formal, others informal:

Sussex University

Sussex University is an internationally recognised ``centre of excellence'' in AI, Natural Language Processing and Cognitive Science, with a proven track-record in practical software development. Within POETIC we aim to undertake research and development in natural language processing and knowledge-based systems, to produce a robust and functional system, and to undertake technology transfer of POETIC research results to the commercial collaborators. Our role is: incremental development (generalisation of language analysis techniques, improving ``robustness'', system monitoring and control, interfaces to dissemination targets), developing support tools (for porting to new sublanguages, for defining new decision procedures, for altering geographical data), moving towards commercial application by ``live'' testing, packaging the system as a ``product'', and supplying user documentation.

Racal Research

Racal Research is the primary research arm of the Racal Group, engaged in digital signal processing, radio communications, telecommunications, CAD, silicon technology, mobile communications (both land-based and satellite), navigation, software development, security, and other technologies vital to Racal. Within POETIC we aim to acquire expertise in AI and Natural Language Processing, to move to a practical working POETIC system and to commercial exploitation as soon as possible, and to identify other applications of this technology. Our role is project management, dissemination of POETIC output, and software and system development.

Automobile Association

The AA is one of the world's largest motorist organisations, and was originally set up to provide road and traffic information. Within POETIC we aim to improve the quality of traffic information given to motorists (data collection is a major problem, and POETIC can be a big help), find new markets for the AA's cartographic database, and get a commercial payback from involvement in POETIC. Our role is to provide cartographic and gazeteer data in electronic form, supply expertise and advice on traffic information broadcasting and ``Intelligent Vehicle-Highway Systems'', and participate in the transition of POETIC from a research and development project into a commercial traffic information broadcasting service.

National Transcommunications Ltd

We are the former Engineering Division of the IBA. The new company is responsible for the transmission of ITV and Channel 4 (S4C in Wales) throughout the United Kingdom and Radio customers include every Independent Radio station plus the majority of Incrementals. With the advent of advanced telecommunications, satellite technology and the application of innovative methods of enhanced picture quality and sound, NTL continues to play a vital and expanding role in broadcast transmission. Within POETIC we aim to encourage exploitation of the Radio Data System (RDS). Our role is to provide RDS expertise, to help in getting access to transmitters and to RDS facilities within the independent broadcasting sector, and to participate in trials of POETIC.

Transport and Road Research Laboratory

TRRL is the research laboratory of the Department of Transport. It is an associate partner in POETIC. TRRL has research interests in new methods of collecting and disseminating traffic information, with a particular interest in their safety and traffic implications. One major objective is to develop strategies for alleviating traffic congestion by giving appropriate advice and information to drivers. Research methods include field trials, behavioural studies and computer simulations. TRRL's role in POETIC is to advise the project on related research elsewhere, particularly that taking place within the European Community's DRIVE project, and assist in contacts with the police and traffic authorities.

Sussex Police and the Metropolitan Police

The police are not collaborators within POETIC, but both Sussex Police and the Metropolitan Police have supplied data which POETIC uses, as well as giving advice about broadcasting of traffic incidents.

see also

Site maintained by: Jonathon Read Disclaimer | Feedback