At Optic Earth we build software that sees through solid rock, creating 3D images of what lies below the Earth’s surface. Our machine learning optimised technology processes robust images 5-10x faster than conventional approaches.
Subsurface imaging is an incredibly complex and time consuming process. With 1000's of parameters to manually set and high computational cost, resulting images can be highly uncertain and subject to bias, leading to an incomplete picture of the subsurface. This adds significant risk to our clients when they are searching for energy, minerals or in renewable applications.
We have developed software that uses Physics Inspired Machine Learning to solve the millions of calculations that are required reconstruct the image of the Earth. As opposed to conventional imaging algorithms that take supercomputers and months of human led parameterisation, our approach can build an image in days, using 1,000-10,000x less compute power. As well as being faster, our machine learned approach can understand and solve more complex physics than possible with conventional techniques. We use this approach to pass on time savings, cost efficiency and improvements in image accuracy to our clients.