How can AI help solve environmental challenges?
How does Sky Wave utilize machine learning to interpret the extensive data collected through drones and remote sensing?
The process of analyzing extensive datasets to gain meaningful insights from remote sensing can be daunting. It can often involve spanning tens to hundreds of gigabytes or even terabytes for relatively small areas. Trying to sift through different data sets to get meaningful answers can be challenging, but remote sensing makes this process much more fluid.
Remote sensing and machine learning are closely intertwined in this process. While machine learning works to process large datasets, the primary challenge lies in effectively navigating through this vast amount of data.
However, machine learning also serves as a critical tool that facilitates efficient analysis and extraction of valuable insights, unlocking the potential of these datasets.
How do we construct a computer model that understands these real-world applications?
In some cases, we use existing computer models as a starting point, but sometimes they’re not built for the right scale or the right environment. So, at times, we need to create new models from scratch. Either way, we utilize real-world, site-specific data collected on the ground for training and validating these models.
Validation is critical to ensure that the model effectively addresses the unique characteristics of the site. We always want to validate and make sure that whatever model we're using or building answers the questions for that site specifically. These models can only predict what they've been trained to predict, so if they haven't seen a problem before, they can't answer it. Some of the sites we work on are very unique and because we're in the real-world environment, these characteristics can change really quickly across geographies.
Which is the recent project carried out by our team where Sky Wave has revolutionized the typical workflow?
Our recent project, which Sky Wave has significantly transformed, involves predicting wetlands using LIDAR data and multispectral imagery. In previous times, conducting fieldwork for such projects would take multiple weeks to cover thousands of acres. However, with Sky Wave's capabilities, we can now create and run models within a few hours. This reduction in processing time has revolutionized our approach, allowing us to expand our project scope to new geographies more efficiently.
What are the myths or misconceptions about using AI for environmental projects?
One main misconception about AI is that it can handle everything and replace human involvement entirely. However, our models are heavily expert-centered and human-centered. They rely on human expertise and insights, emphasizing the importance of people, creativity, and diverse experiences. The collective input of individuals with varied backgrounds and ideas truly enhances the effectiveness of these models.
How has machine learning improved FEMA's assessment of post-code update constructions in the southeastern United States?
A major aspect of FEMA’s analysis involves examining the impact of housing codes and new urban development regulations on building resilience during a disaster. So, in the face of a disaster, their teams assess the performance of structures constructed after these code updates.
Previously, this process involved manually sifting through deed records or toggling imagery to identify changes. However, by leveraging an existing machine learning model and adjusting parameters and input dates, they can quickly flag regions across the southeastern United States where buildings had been constructed post-code updates. This process, which used to take considerable time, now only requires minutes, streamlining FEMA's assessment efforts.
All of the models that we build are very human-centered and so it's the people, their creativity, diverse experience, backgrounds, and ideas that bring the most out of these models
Can you mention an example where the Sky Wave has boosted resiliency and helped to restore the environment?
One of the examples of Sky Wave where it has boosted resiliency and restored the environment is when a bridge was relocated. Sky Wave helped to monitor the return of marshland vegetation to its natural habitat and prevent erosion after the relocation. The use of multispectral drones helped us to quantify the above-ground biomass of the growing plants. By tracking changes in biomass over time, we can assess the effectiveness of restoration efforts and ensure resilience is being promoted as anticipated.
What inquiries typically arise from clients and others when advocating for this technology and how do you handle them?
Usually, people underestimate how machine learning can benefit them. They are more focused on their immediate problem without realizing the potential solutions the technology can offer in the long run. We aim to bridge this gap by understanding their problem thoroughly. Then, we brainstorm available data and how it can be used to address the issue. This approach encourages people to ask more comprehensive and insightful questions, shifting their mindset toward problem-solving methodologies.
What are the types of projects where Sky Wave technology can be beneficial?
Sky Wave technology can be applied to everything visible on the Earth's surface. The sensors can detect more than the human eye, and with tools like LIDAR, they can even penetrate below the canopy to reveal hidden features. For exploring what lies beneath the surface, the technology is evolving, and the issues may require more nuanced approaches but they still are understandable to us and can be addressed effectively.
