As faithful readers of TechWatch (love you, Mom) may know, the rollout of many companies\u2019 ambitious drone delivery services has not gone as quickly as promised. Despite recent signs of progress in Australia and the United States\u2014not to mention clever ideas for burger deliveries to cars stuck in traffic\u2014drone delivery remains a long way from becoming a viable option in the vast majority of use cases. And the problem affects many areas of drone usage, not just the heavily hyped drone delivery applications.\nAccording to Grace McKenzie, director of operations and controller at Iris Automation, one key restriction to economically viable drone deliveries is that the \u201cskies are not safe enough for many drone use cases.\u201d\nSpeaking at a recent SF New Tech \u201cInternet of Everything\u201d event in San Francisco, McKenzie said fear of collisions with manned aircraft is the big reason why the Federal Aviation Association (FAA) and international regulators typically prohibit drones from flying beyond the line of the sight of the remote pilot. Obviously, she added, that restriction greatly constrains where and how drones can make deliveries and is working to keep the market from growing test and pilot programs into full-scale commercial adoption.\n\nDetect and avoid technology is critical\nIris Automation, not surprisingly, is in the business of creating workable collision avoidance systems for drones in an attempt to solve this issue. Variously called \u201cdetect and avoid\u201d or \u201csense and avoid\u201d technologies, these automated solutions are required for \u201cbeyond visual line of sight\u201d (BVLOS) drone operations. There are multiple issues in play.\nAs explained on Iris\u2019 website, \u201cDrone pilots are skilled aviators, but even they struggle to see and avoid obstacles and aircraft when operating drones at extended range [and] no pilot on board means low situational awareness. This risk is huge, and the potential conflicts can be extremely dangerous.\u201d\nAs \u201ca software company with a hardware problem,\u201d McKenzie said, Iris\u2019 systems use artificial intelligence (AI), machine learning, computer vision, and IoT connectivity to identify and focus on the \u201csmall group of pixels that could be a risk.\u201d Working together, those technologies are creating an \u201cexponential curve\u201d in detect-and-avoid technology improvements, she added. The result? Drones that \u201csee better than a human pilot,\u201d she claimed.\nBigger market and new use cases for drones\nIt\u2019s hardly an academic issue. \u201cNot being able to show adequate mitigation of operational risk means regulators are forced to limit drone uses and applications to\u00a0closed environments,\u201d the company says.\nSolving this problem would open up a wide range of industrial and commercial applications for drones. Far beyond delivering burritos, McKenzie said that with confidence in drone \u201csense and avoid\u201d capabilities, drones could be used for all kinds of aerial data gathering, from inspecting hydro-electric dams, power lines, and railways to surveying crops to fighting forest fires and conducting search-and-rescue operations.