Annual Visual AI Survey on the London Principal Road Network

The contract has been awarded to a service provider to provide an annual Artificial Intelligence-Based Visual Carriageway Condition survey (Visual AI survey) to identify and support asset investment decisions on London’s Borough Principal Road Network (BPRN). The condition data collected in the Visual AI survey will be used to move towards a robust approach for funding allocation between the London Highways Authorities to ensure value-driven scheme selection, vital to achieve the best value for money for TfL, the London Boroughs and LoHEG (London Highways Engineers Group Framework).

The service requirement may also include the driven collection of the condition data by the Service Provider as well as a Visual AI Survey of the Transport for London Road Network (TLRN) and Borough Non-Principal Road Networks (i.e. B, C and U Roads) with the provision and online storage of the survey data, as optional services.

The total length of the BPRN required to be surveyed in Lane 1 is approximately 2550 lane km (total for both directions). As well as surveying roads in both directions, some multi-lane carriageways will be surveyed more than once in the same direction. The contract will therefore provide the option for Visual AI surveys to be undertaken in the additional lanes of the BPRN multi-lane network sections, as well as the TLRN and non-principal Borough-maintained roads.

The system used to survey and assess road surface condition shall be an artificial intelligence-based system capable of optical recognition of defects and collection of condition data (meaning the presence, type and severity of defects) using a smartphone camera from a moving vehicle traveling at normal highway speeds. Other mechanical means of survey may be additional but must not replace the use of optics.

The Service Provider shall ensure the collected condition data is processed using an artificial intelligence-based, automated methodology at all stages from collection to processing.

The Service Provider shall carry out all necessary validation and processing of the Visual AI Survey data in order to provide the results according to the required formats and conditions as defined in the Specification and accurately fitting against the Authority’s Network in Geographical Information System (GIS) format.

The Service Provider shall ensure the Visual AI Survey system is able to assess the defects listed for bituminous and concrete carriageways in ‘The UKPMS User Manual, Volume Two, Chapter 8: Detailed Visual Inspection’. These include, but are not limited to:

• Cracking
• Chip loss, fretting and surface deterioration
• Subsidence
• Fatting