Ta Xbiex - 01/01/2010 - The proposed project aims to generate new processes and skills in the area of vehicle analysis and automated intelligent image analysis systems, which we are collectively referring to as an “Intelligent Digital Traffic Enforcement System” (IDITES) with the aim of creating an entirely new niche market. Currently, traffic enforcement is a largely manual process, with significant amount of human resources being spent on enforcement and monitoring activities, in both on-street (public) and off-street (private) contexts. It is highly desirable to have an intelligent automated system that is capable of flagging enforcement events automatically, reducing the role of humans to perform the last verification step before enforcement is carried out. Such a system, when widely deployed will lead to a better and fairer enforcement system, while freeing up precious human resources for other activities, such as patrolling neighbourhoods against theft, helping out social communities and maintaining order and giving better service rather than focusing on traffic enforcement almost exclusively. The timing is also opportune as there is currently a need to upgrade old analogue systems with modern digital versions. There is also a drive to have one device performing multiple functions. Brief Points - Safety and enforcement needs should be balanced by needs of motorists - Human wardens and officers should focus less on traffic offences and use the advantage of human contact to improve matters at a more social level – tackling anti-social behaviour, littering, etc. Main Research Objectives - Current image processing algorithms cannot handle the task in road-side equipment – need development of new, more efficient algorithms - Need to improve level of proof to reduce false accusations against drivers and reduce mistakes - Reduce carbon emissions via low-emission zones and automatic detection of particulate emissions - Enabling advanced processing to be done via lower power computing hardware – need research on less power hungry software that can run on low power hardware, for example, solar powered embedded boards Existing Skills Charonite already has considerable expertise in related fields, having developed various applications for use in ANPR based systems, parking management, enforcement and congestion charging. The team envisaged to implement the proposed project will ensure that the project deliverables are of high quality. Screenshot of existing Charonite manual enforcement application New Skills and Knowledge The co-financed activities all form part of the Industrial Research project being proposed in this application. The proposed project will create new skills, processes and knowledge for Charonite that is beyond the current state-of-the art. The expert preliminary study (attached as part of Annex 10) confirms this. The new skills and knowledge that will be created corresponds closely to the project deliverables, and are as follows: - Creation of an IDITES framework – this will serve as a technology validation testbed / platform - Creation of a new class of image processing algorithms – a significant amount of R&D resources will be allocated towards the creation of image processing algorithms that can handle messy real-life imagery from multiple disparate sources and understand complicated scenes that are fed into an Artificial Intelligence system for rule violation detection - Research on using low power parallel processing to make roadside processing feasible – one of the main barriers towards having a practical future IDITES system is the amount of power (both computing power and also actual electrical power for machinery) that is needed to process realtime video feeds. In order to make roadside processing feasible, research is needed on having algorithms and techniques that can run on low power parallel processing architectures. In particular, it is proposed that FPGA techniques are researched as part of the project, something which is novel. - Research on automatic emission detection – vehicle emissions are currently not detected in realtime and linked to a vehicle. The proposed IDITES research will create state-of-the-art techniques that go beyond the most advanced system in the world, the London low emissions zone, which is entirely dependent on a database of vehicles rather than actually processing and interpreting what is going on in the real world. - Research on automatic rule violation detection and presentation – the development of an Artificial Intelligence based rule engine that can analyse and understand scenes in realtime and trigger off violation rules, using positional awareness and a sophisticated optimised schedule.