Computer Vision News - October 2021

20 simply to fly near a population then you have to validate your system with a single point of failure, which has the probability of one divided by 10,000. This is the number of flight hours you have to do, and you have to do a strong validation of your system. This is a big opportunity for us because we know there are many companies who want to do that, but it’s not solved yet. I believe there is a long way to go before we see package delivery outside of a test field, so we want to solve an easier approach first – pipeline inspection where no population is nearby . You still have to show that your system is safe if you want to fly far with the pilot out of the loop. That is called BVLOS – beyond visual line of sight . This is not yet solved for certification for most of the use cases in North America and Europe.” Spleenlab have some very exciting plans for next year and beyond. As a software company, they collaborate with manufacturers to bring AI to their products, and at the beginning of next year will be launching their simple follow me functions up in the air with drone manufacturer Quantum- Systems . By the end of next year, Stefan says he hopes to see Spleenlab’s detect and avoid system in drones. They are also working on automatic inspection of cell towers , with the AI looking for the cell tower, flying a drone around it while collecting inspection data, and then bringing it back. This will save money for customers who want to automate the inspection process. Spleenlab are currently 15 people, and they are hiring. “We are looking for computer vision engineers, deep learning engineers, and PhD applicants with a focus on SLAM perception and sensor processing. Come join us!” Semantic Landing Risk estimation AI Systems for Autonomous Mobility