04 Sep 2019
Artificial Intelligence on the Shrimp Farm: Osmobot
Written by Zach Stein

We have been hard at work on bringing Osmobot, a water quality sensor, and monitoring system built for fish and shrimp ponds, to market. In this post, we’d like to share some of our latest advances and what they mean for our customers. 

Latest Osmobot Update: Using Computer Vision and Artificial Intelligence

For those who have been following our development, you will know that Osmobot’s technology has always been centered around analyzing robust, photochemical sensor patches for each of the core parameters it monitors: dissolved oxygen, pH, and ammonia (NH3). 

Historically, Osmobot has done this using individual photodiodes to track the light coming back from each sensor patch. Now it uses a camera to take a picture of the entire sensor cartridge and our custom artificial intelligence algorithms to analyze the picture and generate a reading. 

More Accurate Readings

What this means is that our readings are more accurate than they have ever been. Our AI algorithms scan the entire image, looking for tiny patterns that humans cannot detect, and generate a reading based upon the algorithm that it has been trained on. 

This much larger amount of data in each reading, allows the AI-based algorithms to more easily distinguish between the noise in a sensor reading (caused by other changes in the environment) and the signal which is caused by a change in a given parameter. 

Ability to Learn and Improve Overtime in Every Pond

What is exciting for us about this is that Osmobot is the first sensor of its kind that can continually improve over time in the pond. The more locations we train the sensor in, the greater our library of data to build increasingly strong and resilient sensing algorithms. 

This not only can extend Osmobot’s lifetime in the water in between cartridge replacements, but also can help it become even more accurate and sensitive to smaller and smaller shifts in the environment in the pond, which can give farmers further and more accurate advanced warnings of potential danger. 

Conclusion

We are going to be field testing this next version of Osmobot shortly. Stay tuned for future updates. The future is bright.