AI and Remote Sensing for Progressive Landscape and Urban Design

AI and Remote Sensing for Progressive Landscape and Urban Design

AI and Remote Sensing for Progressive Landscape and Urban Design News / Publication / 02.03.2021

With Dr Hossein Rizeei 

Artificial intelligence (AI) and Machine Learning (ML) technologies have begun to shape the design industry in all its disciplines, and as cities continue to evolve to adapt to contemporary issues such as climate change, so should our design tools and thought processes.

Dr. Hossein Rizeei, McGregor Coxall’s Geospatial Scientist, has been at the forefront of research dedicated to AI and ML implementation in remote sensing throughout his career. His expertise in the field is recognised through his appointment as Topic Editor for the MDPI Remote Sensing Journal and as a Guest Editor for the Special Issue \"Advanced Application of Artificial Intelligence and Machine Vision in Remote Sensing\"(ISSN 2072-4292), and recently member of the Editors Board for the Natural Resources Informatics - Spatial Issue of the Remote Sensing Journal.

Backed by over 16 years of research including 32 publications, Dr. Rizeei has created and implemented these methodologies and technologies to help inform design solutions for our landscape, urbanism and environmental teams across the globe. This approach ensures that our design solutions are founded in solid, evidence based data and allows us to achieve more sustainable and resilient outcomes.

So, how exactly is this technology being utilised?

Greater access to reliable data

Hossein uses remotely sensed aerial or satellite data to extract and process information that can be used to inform design. Examples of extracted information includes canopy cover, tree height, solar radiation and building height to name a few. The extraction and processing process (known as object detection) is traditionally performed manually, making it a user-subjective exercise that can be both confounded by biases and tedious in its approach.

Hossein uses Machine Vision (MV) to both improve the efficiency of this process, and to eliminate human error and biases by applying his custom-built, Geospatial-AI models on the data that can be objectively validated and replicated.

Technically speaking, conventional training models with remotely sensed data are performed manually, making it is a user-subjective exercise that’s both unclear and tedious in its approach.

Rizeei, a member of the McGregor Coxall team since early 2020, MV intends to be rid of these uncertainties by establishing consistent, valid, and reliable methods.

“The AI technology attempts to leverage these current geospatial systems, including remote sensing, in a novel robust way to provide an automatic inspection workflow. This includes image acquisition from the sensor, digital image pre-processing, training, and testing techniques, validation and knowledge extraction.

“Through a combination of high-powered graphics processing unites and deep learning AI, we are able to integrate architectural and geospatial analytics software to assess these outcomes.”

Complex data analysis

Hossein has created a framework that combines several technical tools and methods to harness different types of raw environmental and social data to tailor our understanding of the existing conditions.

This allows him to predict possible outcomes to a project site as a result of proposed works and present the holistic data in a palatable way for the client. As a design team, we use these methods to undertake wide reaching multi-criteria assessment (MCA) to inform the performance of design scenarios.

“Through my models, we are able to assess design options against a very wide range of triple bottom line performance criteria of, for example, a new urban development. Similarly, we are able to calculate the suitability of areas for certain development types based on customised objectives and KPIs.

“For the first time within the industry we’re seeing design advancements by AI and ML having a profound impact on the way we shape our cities and communities,” said Rizeei.

Predictive analysis

“The tools that I have built predict environmental, agricultural, socioeconomic and project disaster forecasts using visual and numerical data sourced from multi-sensor remote sensing, geo-statistics, ArcGIS, 2D hydraulic techniques, integrated parametric methods, optimised time-series hydrological models, machine learning models, and data-driven logistic regression models,” said Dr Rizeei.

One such tool is forecasting and time series analysis: a primary instrument used in various environmental applications such as urban planning, hydrology, and watershed management.

Hossein explains its application on the Sunshine Energy Park project.

“Environmental events aren’t straight forward phenomena and are subject to many conditioning elements. We proposed a full package Geospatial-AI model to first detect the existing condition and real changes of land use over time, and then to forecast of the future scenario based on detected trend function and zone planning rules using statistical and ML approaches.”

What kind of impact does this have?

While designers often design using their professional experience, their research into precedents and their inherent design sensibility, Hossein approaches design issues from a data analysis perspective, analysing layers of site-specific data to compute thoughtful and complete design options.

When these two aspects converge as part of McGregor Coxall’s multi-disciplinary approach, we are able to create design outcomes that are both objectively sound across a wide range of critical performance indicators and possess the beautiful, compelling aesthetic – the less ‘calculatable’ values of design that McGregor Coxall’s award winning designers bring.

“Architects continue to manage a project’s imagination and vision and are now able to deliver them with more clarity and precision thanks to AI and ML algorithms,” said Dr Rizeei.

Hossein’s work allows key stakeholders in the design process to be more involved and informed by clearly presented new data and information that ultimately impacts decisions and methodologies surrounding the design of a built environment.

“Currently, we are harnessing rapidly growing visualisation and presentation technologies such as ArcGIS and Power BI interactive dashboards to improve how we communicate our analysis and findings to stakeholders. This provides them with added value of being able to see, manipulate and analyse new data that we have created without needing to be experts in specialist software.”

What’s next in the field of AI and design?

Dr. Rizeei says:

“As we know, modernization is a nonstop development. Moving from manual sketching to computers didn’t occur overnight, though. Adopting new technologies has always been a challenge as the fear of being replaced makes it difficult to see what this opportunity can bring for human beings.

“In the future, Geospatial-AI will change the way we look at our earth. It will enhance our ability to observe and interpret data, particularly through the adoption of a single platform that supports geospatial data in all its forms.

“The processing of national data would be a simpler procedure and will ultimately save firms and clients time and cloud storage. Additionally, critical GIS voluntary data will be frequently assessed to create a rich, reliable and up-to-date spatial knowledge database which can used for all scales of urban planning.”

About Dr Hossein Rizeei

Hossein is a highly qualified GIS scientist with more than 16 years’ experience as a data scientist, geospatial modeller, geospatial data engineer and environmental project manager. He has designed, developed, and tested several geospatial-based methodologies to assess urban planning, natural hazards, green infrastructure metrics using GIS, remote sensing, artificial intelligence, and statistical techniques.

Hossein’s repertoire in delivering informed solutions includes the employment of machine learning models in geospatial systems, geospatial 3D analysis, geospatial database querying, Web GIS, airborne/spaceborne image processing, feature extraction, time serious analysis in forecasting modelling and domain adaptation in various environmental applications.

Throughout his professional life, Hossein has also published several ISI papers, book chapters and conference scientific articles in these fields, while more recently delivering remote sensing assessments to extract the tree canopy coverage associated with height from Lidar point clouds and informing best practice in urban greening factor metrics.