Torsten Merz is a senior research scientist and engineer at the Commonwealth Scientific and Industrial Research Organisation’s Autonomous Systems Laboratory in Australia. He has worked on several projects using UAS in agricultural applications for low-altitude remote sensing. Increasing Human Potential recently had the chance to catch up with Merz to learn more about his work.
You’re currently doing some exciting work with CSIRO regarding low-altitude remote sensing tasks in agriculture and biosecurity in Australia. Can you tell us about it and what you hope your research will accomplish?
I have worked on several projects requiring dependable autonomous flight of unmanned helicopters for low-altitude remote sensing. In one of the projects we developed an aircraft system for crop monitoring in plant breeding trials.
In the latest project, which was part of a larger collaborative project named ResQu, we developed a system to find an invasive plant that threatens the Australian rainforest.
For the first application, we fitted a helicopter with a multispectral camera system, allowing plant breeders to efficiently gain important information from different varieties of crops at different growth stages. As some of the field trials required longer flight times and the sensor payload was heavy, we chose a conventional single-rotor helicopter with petrol engine for the task. One of the main challenges was to design the system in a way that it can be safely operated by the end user directly without pilot training. The same system could also be utilized by farmers to monitor their crops, helping them to decide where to irrigate or where to apply herbicides or pesticides.
For the biosecurity application we fitted the same helicopter type with a sweeping camera system, allowing it to capture wide-angle high-resolution images. The main challenges were the mountainous environment the aircraft had to operate in, limited communication with the helicopter and, again, safe operations without pilot training. Apart from dependable autonomous flight, accurate georeferencing of images and high image quality was required. All images were recorded on board the aircraft and processed after the flight.
Apart from providing cost-effective and better solutions to real-world problems, I want to demonstrate with the developed prototypes that it’s possible to safely deploy complex autonomous aircraft for specific tasks requiring minimal human interaction and training.
Researchers and industry are finding new ways to apply UAS almost every day. What drew you to your current work regarding agriculture and biosecurity?
We have been working on the automation of unmanned helicopters for research purpose for quite some time. Apart from being research platforms, rotorcraft are also a good choice for many low-altitude remote-sensing applications as they allow accurate positioning of sensors in 3-D space at almost any time, they don’t require a minimum speed to fly and they can easily be deployed almost anywhere. This was realized by our crop scientists, and we started to collaborate, which eventually lead to the crop monitoring project. While we tested the systems at different locations, we also spoke to farmers who showed interest in the technology.
In biosecurity, aerial surveys are often conducted with manned helicopters. The patterns the aircraft must fly are tedious for the pilot, and low-altitude flight has also its risks. The idea was to get the humans out of the aircraft and turn the aircraft into a robot. Robots are much better suited for repetitive tasks that require precise flying. We also received state government funding for addressing a problem which was relevant to Queensland. Miconia weed spotting seemed to be ideal as the flights are conducted over unpopulated terrain in uncontrolled airspace with practically no other air traffic at the height we operate. Moreover, there had already been an accident during a manned helicopter survey, so there has been a demand for a safer solution.
What benefits do UAS provide in conducting your research, compared to their manned counterparts?
I’m interested in the development of complex robotic systems requiring dependable autonomy. Unmanned aircraft, which need to operate beyond radio range or which are deployed by users without special training, require dependable autonomy. Some system components can be developed and tested using manned aircraft. Moreover, optionally piloted flight can be a good way to overcome regulatory restrictions.
However, at some stage control loops must be closed, and embodiment becomes important. If the robot is a full-size aircraft, the test pilot can stay on board for a quite a while, but currently our target aircraft are much smaller. Moreover, unmanned aircraft operations are cheaper for us, and we can deploy our helicopters nearly anywhere at practically any time, and modifications to the aircraft don’t require regulatory approval. We also do some research which requires persistent flight capability, which we currently only have for small unmanned aircraft.
During your time leading the development of system components enabling autonomous flight services at CSIRO, how has UAS technology evolved? What have you found to be their most useful application?
Over time, the focus has shifted from larger, mechanically complex unmanned aircraft which required good pilots to affordable, small, electric multicopters and foam planes with attitude stabilizers or autopilots which can be operated by practically everyone. Unmanned aircraft can now also be bought as complete package including a lightweight sensor payload, a ground station and mission planning software. However, these aircraft are not suitable for all types of applications. For high payload, long endurance or long-range missions the larger more complex aircraft are required ,and there are practically no cost-effective systems available. Moreover, most commercially available systems don’t offer important capabilities for autonomous flight such as obstacle avoidance, health monitoring, mission planning incorporating environmental, and regulatory constraints and decision making. In research, however, these problems are currently addressed. I think in industry that hasn’t been a priority for such systems, as for many applications, small aircraft do the job and it’s not a problem or even desired to have a human pilot. Moreover, we don’t have regulations yet allowing autonomous flight.
Apart from the applications we are addressing, successful civilian applications for smaller remotely piloted aircraft, with attitude stabilizers or autopilot support, seem to be aerial photography and filming, close-range infrastructure inspection, damage and accident scene assessments, and mapping.
Given that you’ve spent time in Australia, Sweden, and Germany and have a strong international perspective, what are some exciting and innovative uses of UAS you’ve seen around the world? Where would you like to see the technology go next?
I’ve attended several demonstrations and flight tests of unmanned aircraft around the world, but I haven’t seen the technology much deployed in civilian real-world applications. I heard about new exciting applications, such as quadcopters for medicine deliveries to the island of Juist in Germany. I also once saw a swarm of quadcopters used for an artistic display by a group of people from Austria, which could be an interesting alternative to fireworks.
I would like to see more dependable unmanned aircraft that can be operated safely without pilot training for applications that are tedious or too dangerous for humans or which would not even be possible with conventional aircraft. On the regulatory side, I would like to see the authorities to focus more on the automation aspects and the dependability of the aircraft and what they are deployed for.
Sometimes the work that’s done in academic contexts has a hard time transferring over to industry application. In your opinion, how can both academia and industry better collaborate in order to further develop unmanned technology?
One problem is that academia typically focuses on lower technology readiness levels whereas industry typically picks up technology at higher TRLs. So there is a gap which organizations such as the CSIRO are aiming to fill.
Furthermore, industry should talk more to academia about their problems and provide funding for industry research. Academia on the other hand should listen more to the demands in industry. In practice, however, researchers often address problems they find interesting rather than what industry requires. Moreover, industry and academia often work on problems in parallel without knowing about each other, which is partly due to commercial confidentiality issues, but also due to lack of communication between the two worlds.
What does that future hold for you, CSIRO and unmanned systems? Do you have any exciting projects planned? What would you like to do next?
I’m generally interested in developing dependable autonomous systems for real-world applications where dependable autonomy is not a desired, but an essential requirement. In the last years, I’ve been dealing mostly with unmanned helicopters, but I would like to work on a wider variety of vehicles. CSIRO has been working on unmanned systems on the ground, in the air, and in the water and hopefully will continue doing so. The unmanned aircraft market, however, is rapidly growing and research funding is more likely to become available in this domain. I’m currently trying to secure funding from industry for the development of autonomy technology for larger aircraft operating beyond line of sight of the operator.
In general, I would like to see unmanned systems technology being utilized to address the problems we are currently facing in the world, and I would like to contribute to make it happen. However, we have to be careful to actually solve problems and not create even worse problems with the technology we develop.