FrontierSI have released a comprehensive report that highlights the gaps in understanding between UAV data providers and the needs of the geospatial industry in creating data products that meet project requirements.

Based on extensive research and a survey of over 135 responses from 80 organisations, FrontierSI’s ‘UAV data acquisition in Australia and New Zealand user needs’ report identifies the user needs of nearly 60 distinct applications across 12 industry sectors — and describes the individual technical requirements of these in detail.

Seeking to tackle the ongoing issue of low data quality provided by some data providers and resulting costs and project setbacks for users, the researchers found clear deficiencies in established and documented standards for UAV-borne data capture, and widely varying quality assurance processes.

Image provided by PRNewsFoto/Insitu, Inc.

Findings

The research found that optical cameras were the most common payload for data capture, most often used to acquire photographic imagery of infrastructure and landscape features, noting that infrared, multispectral and hyperspectral sensing was often required in agriculture, forestry and asset inspection applications.

Accuracy and image quality are the standout issues identified by surveyed users of UAV-acquired data, with around half of respondents naming accuracy as a problem, including lack of positioning accuracy, incomplete data, lack of georeferencing, gaps in coverage and incorrect units.

Stay up to date by getting stories like this delivered to your mailbox.
Sign up to receive our free weekly Spatial Source newsletter.

FrontierSI attributes this in part to a lack of published guidelines, with most users unaware of formal standards, and only 12 percent of respondents using an existing standard for capture, noting that standards for manned aircraft capture are not transmissible to UAV-based practice, but that data providers revert to these in some cases.

Image quality was an issue for just under a third of respondents, with the researchers noting that needs differ wildly between applications — resolution may be critical in some, while sharpness, colour balance and haze-free images may be more important characteristics in others.

Despite typically being far cheaper and easier to carry out than manned aerial surveys, UAV-based capture operations face their own challenges — among them weather factors, equipment malfunction, site access issues, data processing and software problems, and specification-related issues resulting in inappropriate products being delivered.

Image provided by Aerometrex.

Conclusions and next steps

FrontierSI highlighted over 40 quality assurance checks aimed at assessing quality of and fitnes-for-purpose for UAV-derived data — along with a ‘clear need’ for quality assurance tools to assist in determining appropriate data specifications.

The researchers hold that specifications should be determined on a per-project basis, and identified the need for a standardised workflow to guide the data procurement process.

The largest gap found, however — was for quality assurance tools and processes. While some processing suites like Pix4D and Photoscan can automate some level of QA and reporting needs post-capture, many of the checks recommended by the report cannot be performed simply or by non-technical users.

The outcomes of FrontierSI’s research will inform development of a ‘QA4UAV’ app, which will aim to provide a workflow for users to:

• Determine their data capture specifications;
• Ensure the quality of captured data in the field;
• Perform a rigorous QA on the acquired data in the
  office; and
• Work collaboratively with providers in the delivery of
  their projects.

You can access FrontierSI’s report, including the technical requirements for their identified applications here [PDF].

Stay up to date by getting stories like this delivered to your mailbox.
Sign up to receive our free weekly Spatial Source newsletter.