Xtend raises $40M to allow human commanders to guide autonomous AI drone missions – Business

Xtend, the developer of the artificial intelligence-driven operating system for drones and robots, today announced it has raised $40 million in a Series B funding round led by Chartered Group with participation from Clal-Tech.

The company’s AI-powered drone operating system allows human operators to perform demanding and dynamic missions with drones and robots in the field and indoors with minimal training. Drones under the control of XOS can receive simple commands and gestures to translate them into intuitive actions the same way another person in the field might, and deal with real-time changes in the environment.

Xtend co-founder and Chief Experience Officer Matteo Shapira told SiliconANGLE in an interview that for drones to assist in field operations, such as with police, firefighters or soldiers, drones needed more complex capabilities than when they were being used to hover above a scene and provide an eye in the sky. 

Early on, the team created the neural network algorithms that allow the robots to navigate in complex environments, and overcome obstacles to complete tasks, such as tracking subjects and identifying objects to perform search and rescue or follow targets.

What the team is doing now is building what Shapira called “common sense practical autonomy” for drones and other robots.

“We assume that there’s always a person in the middle who is a professional human being, who has the capability and the experience to instruct the robot or a team of robots,” Shapira said. “We’re not trying to develop a drone, or robot, that’s 100% autonomous that you’re just sending on a mission.”

Essentially, Xtend is building a technology layer using the already existing XOS drone operating system that will allow drones and robots to use the complex guidance already built into them to complete missions using neural networks, but under the command of a human.

A search and rescue drone operator could have multiple robots enter an earthquake-damaged building and ask one to scan the first floor and have it report back if it finds anything. Then have the second one patrol the second floor and have a drone hover above. 

Using XOS, the operator can easily switch views between the different drones while they’re in operation and see through their sensors or receive alerts when they detect anomalies as they traverse the environment on their mission. This gives the commander a chance to revise their parameters on the fly as priorities change on the ground.

“That gives you a lot of operational advantage,” Shapira said. “The technology that we’re releasing soon, you can do it without even manually connecting to the drone at all.”

Users will get access to both gesture controls and menu options to control either individual robots and drones in their fleets or entire groups. This will allow them to dive in and micromanage, should they choose, or see everything from above. Shapira added that the team is working on plans to involve three-dimensional mapping capabilities and a next-generation with large language model chatbot so that voice commands can be given to the drones as well.

The drones themselves learn as they operate in the field, just like humans do. “We’re essentially teaching the drones to think like a human operator would, based on all the hundreds of missions, and the training data that we have,” Shapira said. “Scanning a building traditionally would be like a 2D or a 3D maze AI algorithm, so to speak. But what we’re doing, it tries to mimic what an actual human would do in that situation when given that specific command.”

Shapira said that the company intends to use the money to help centralize the technology stack into a “proficient operating system,” that will give drone and robotics companies all the capabilities they need in one place. This will not just include the AI autonomy stack that is being released today, but also third-party app support and other features that will enable drone operation companies to build on top of drone platforms.

Image: Xtend