Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationBright Pick Auto Picker5,000,000 picks used for training
Bright Pick trained AI algorithms on 5 million picks to enable a mobile manipulator to sort and fit goods into bins with high accuracy, allowing it to handle a wide range of items like groceries and electronics without human intervention.
“They've been training the algorithms for auto picker on 5000000 picks today, and it's constantly improving as we go along, and I think that's starting the key … to see AI used in a way to break through that barrier of saying, oh, how do we pick a wide range of goods? How do we improve the accuracy to 99% or more?”
Engineering & simulationElectric Sheep Foundation Model
Electric Sheep acquires commercial landscaping businesses to gather real-world data, which is then used to inform and improve their robot's foundation model before widespread deployment.
“And then the cherry on top is that they also get to acquire all of the data that those companies bring in. So they're making their robots better before they're even deploying them using real world data.”
Engineering & simulationOptaran MyNetwork
Optaran developed a machine vision model inspired by insect brains that processes data from low-cost sensors to enable autonomous mobile robots to navigate complex environments with mirrors and varying light conditions, overcoming issues with transparency and reflectivity.
“The new campaigns company has been inspired by recent studies of how inside frames work. … a product that can take data from low cost sensors. … and it tanks that data for localization, obstacle and detection and navigation to do things that other AI models can't.”