UST034

23 circumvent that, but I also optimised the Stork’s parafoil’s designs with a laminar flow boundary-layer model afterwards to fit the speeds, ranges and other mission requirements we have.” Animal Dynamics has developed two Stork models, for last-mile deliveries in dynamic and potentially dangerous conditions for defence, disaster relief and medevac applications. One is the STM, which can carry payloads of up to 120 kg (with its MTOW yet to be determined), while the ST25 can fly for 50 km or more, and carry up to 10 kg of payload, with a MTOW of 25 kg. “The STM will be ready for certification and field trials next year, and we expect to deliver on the first substantial orders in 2023,” Prof Thomas says. “The proof-of- concept version flew in 2019, and the ST25 has already been field-tested and sold in pre-certification trial units.” Both models are designed to maximise payload capacity and endurance while minimising costs from operations and maintenance, which Animal Dynamics predicts to be the prime future driver of autonomous aerial deliveries, especially in disaster relief or rural areas without a road infrastructure. To deliver urgent payloads in a timely fashion, the UAVs are being trained using intelligent feature recognition to autonomously pick out optimal landing sites near the assigned endpoints of their missions. Prof Thomas estimates that a tennis court-sized flat area is sufficient for the craft to land and take off. “The vast majority of UAVs in the Storks’ weight classes don’t actually have real autonomy – they follow GPS waypoints or carry out a ‘follow-me’ mission using vehicles on the ground. They can compensate for wind and detect large obstacles, but many have problems with rerouting and similar kinds of second- level intelligence problems,” he notes. “If we’re delivering emergency fuel, food, water or something else, however, we have to be able to identify multiple potential landing sites on-the-fly, and dodge any power lines, trees or other obstacles to land there without delay. “Our lead computer vision person actually came to us from the autonomous car space. She was working on convolutional neural networks [CNNs] for obscuration problems such as when one vehicle disappears behind another, and now she’s making CNNs for the data segmentation in spotting a landing area. “That is quite a bit more straightforward than training self-driving cars, and can use light-duty processors. We’re using Nvidia Jetsons because they’ve become the standard, but we could switch to less powerful ones – a Raspberry Pi for example would probably work.” Conclusion With different Stork vehicles now being worked on, Prof Thomas anticipates development of the Skeeter and Malolo to accelerate thereafter, and sees many more directions in which bio-inspired technologies could be used. “Whether you start with a fish, bird or insect, you can see common advantages that evolution arrives at, which are worth exploring for unmanned systems; things like streamlined tuna-like body shapes for UAVs and UUVs. And carcinisation – becoming increasingly crab-like, with multiple degree-of-freedom manipulators and flexible alignment – is becoming increasingly common for ROV-type UUVs and some UGVs. “When you look across the animal kingdom, there really is so much low- hanging fruit that nature has to offer in terms of what we’re poised to adopt for advancing robotics in the future.” Unmanned Systems Technology | October/November 2020 Prof Adrian Thomas is the chief science officer at Animal Dynamics, where he heads the r&d for a series of projects aimed at improving the efficiency, stability and intelligence of unmanned systems by adopting and optimising bio-inspired engineering techniques. His childhood was spent in Oxford county, England, where his interests in marine and avian wildlife and designing competitive model aircraft led to a career in bio-inspired engineering research, stretching from Britain to Sweden and Australia. He later became a fellow at Oxford University in 1998 and a professor of biomechanics there in 2006. He met entrepreneur Alex Caccia in 2013, and in 2015 the two founded Animal Dynamics in order to mature and productionise some of Thomas’ designs for unmanned underwater vehicles and aircraft. The company’s Stork STM UAV is due to complete certification later this year. In addition to his academic and business career, Prof Thomas is also an accomplished paraglider, having won the British National Championships in 2006, 2009, 2013 and 2014 – wins aided by (and adding to) his aeronautical engineering training and experience. Adrian Thomas