- Unmanned submersible autonomously gathers underwater data in Lake Travis test
- Led by former Navy SEALs, Terradepth combines machine learning, edge computing and novel energy recharge technology to collect data at scale
- Attains measurable, important progress towards long-term vision of increasing ocean knowledge for science and humanity
Terradepth, a disruptor in maritime data collection and use, today announced the successful completion of its Phase 1 trials. The test was executed at Lake Travis in Travis County, Texas. The Phase 1 test results conclusively demonstrate that the company’s unmanned submersible could collect and process underwater data, understand features of import, and automatically retask itself with no human intervention.
Terradepth’s mission is to increase ocean knowledge through autonomous, high-resolution, scalable data collection and a radically improved data experience. The company is applying autonomous robotics, AI/ML, and the latest software concepts and methodologies to create the world’s first deep ocean data-as-a-service business.
The initial trial is one of multiple tests that will prove out Terradepth’s hybrid ocean data collection submersible, which will operate autonomously to collect an ocean data repository of unprecedented scale.
- “Deep ocean data promises to enlighten and advance us on everything from the understanding of flora and fauna to weather to how the world works,” said Judson Kauffman, co-founder and co-CEO of Terradepth.Succeeded in the overall mission of demonstrating basic, end-to-end in situ automated data processing in a known submerged environment;
- Ran a proprietary machine learning model algorithm on the robot to autonomously detect objects of interest;
- Proved Terradepth’s proprietary data extraction capability for preparing sonar data for onboard processing;
- Created and demonstrated an end-to-end autonomous onboard data processing pipeline to enable automatic target recognition and follow-on autonomous retasking – removing the human from the data interpretation and retasking functions;
- Test parameters:
- Functionality of computing hardware inside a pressure vessel mounted inside the robot, submerged in water;
- End-to-end functionality of the data processing pipeline on the robot;
- Capability of the data processing pipeline to send “snippets” of information to humans for quality assurance and objects of interest;
- Accuracy of the machine learning model’s inferences contrasted with known subsurface objects of interest.
“The success of our first trial is an important first step towards democratizing ocean data, and is another important step toward our goal of sharing information that can help to conserve and protect 98.5% of Earth’s livable space — the ocean,” said Joe Wolfel, co-founder and co-CEO at Terradepth.