skip to Main Content

IN DETAIL: Anduril launches extended Lattice OS

Dr Shane Arnott at Anduril’s facility in Sydney. He is standing next to the Dive-LD Autonomous Underwater Vessel (AUV) which is equipped with the same Lattice OS as the Ghost Shark Extra Large AUV (XL-AUV) the company is developing for the RAN; now the company has launched Lattice for Mission Autonomy (top). Images: Anduril

Anduril has launched an extension of its AI-powered Lattice OS which is designed to enable a single operator to manage multiple autonomous systems. And despite the new system, Lattice for Mission Autonomy, only making its public debut at a defence show in the US at the beginning of May, it is already available in Australia.

Described by the company as “a hardware-agnostic, end-to-end software platform” that operates across multiple domains, Lattice for Mission Autonomy is designed to allow humans to employ autonomous systems much more effectively, extending reach, capabilities and situational awareness while enabling warfighters to make better decisions, faster. The system could become a key part of a force-level Command, Control Communications and Information (C3I) system, automating much of a commanders’ and headquarters’ job.

“This [new] Operating System (OS) was created really to help the human deal with a much more complex battlespace,” says Dr Shane Arnott, Anduril’s Senior Vice President Engineering. It is designed to sense the vehicle’s environment, provide a framework within which humans can make some high-level decisions and then enable the vehicle – or equipment, or system – to take the appropriate action.

“It sits above teams, and teams of teams,” says Arnott. In the American parlance, he says, you can tell it to score a touch-down against a particular team. You don’t need to say more, still less tell it what plays to use. It aims to deliver the outcome with whatever assets you bring to the mission.

Lattice for Mission Autonomy is designed to understand mission intent, says Anduril. This enables the AI system to deliver the operator’s desired outcomes in complex, multi-player missions. In the military context, he says, you might have a mixed force of autonomous fighter aircraft, some surveillance aircraft and some jammers – Lattice for Mission Control will work out the best ‘play’ for a given situation and stated outcome and it will dynamically re-task the players to deliver that outcome.

The new OS is an extension of the Lattice OS used on the Ghost Shark Extra Large Autonomous Undersea Vessel (XL-AUV) that Anduril Australia is developing in partnership with the RAN and DSTG at a secret location on Sydney Harbour, says Arnott.

So, to use Ghost Shark as an example, using the new system a team of these vessels could be tasked simply with an action and then left to get on with it. However, they would also check in regularly with the human force manager, says Arnott, especially when they encounter an unanticipated event that requires some sort of human input. In the normal course of events they’d move checkpoint by checkpoint and conduct proper phasing as part of their execution of the commander’s intent.

He won’t say when we might expect to see either Lattice OS or Lattice for Mission Autonomy in service. Arnott tells EX2 that they have already been used on exercises in Australia that he’s not at liberty to talk about; the company is building up Ghost Shark, he says, and has other Australian programs incorporating Lattice and Lattice for Mission Autonomy that the company can’t talk about at present.

“At some stage we’ll get the opportunity to talk about that publicly,” he says, “But we’re not there yet.”

AI enabling C2 and C3I

Anduril’s strongest suit is its software. Much of its hardware is developed by partners and recently acquired subsidiaries but the essential link between them all is the common AI software that enables autonomous operations, both individually and in teams. This is why Anduril has focussed on developing the AI-enabled Lattice OS and Lattice for Mission Control: these systems are hardware-agnostic and can be applied to any fleet or force as part of a commander’s C3I system.

A joint force commander has to manage and control multiple different sensors and effectors. The more complex the operation, the more difficult this management and control task becomes; the bigger the headquarters becomes; and therefore the more constrained the information flows within that headquarters. An AI-enabled C2 system that enables and manages the autonomous operations of the force’s sensors and effectors will simplify and streamline an important part of the commander’s job, leaving him or her free to think, plan and exercise judgement.

The software-driven Lattice for Mission Control management terminal would be identical to that currently used for the Ghost Shark and for Anduril’s  counter-Uncrewed Air System (UAS), Tower surveillance and ALTIUS UAS that is reportedly now being used in Ukraine (Arnott is unable to comment). And it could be situated on a ship, in a ground vehicle, on land somewhere or in a Wedgetail AEW&C aircraft.

A potentially significant challenge will be to identify within a joint force who is the most appropriate commander/manager of the autonomous fleet. This would likely be a member of the force commander’s staff, but if circumstances change – if an airborne Command Post is required to pull back from a threat or if a land-based commander is forced to make a sudden move, or if the command team simply gets overloaded – management of the autonomous network, or even just a part of it, could be handed dynamically from one node to another as required. And if an effector goes beyond communications range, for example, the system would task another platform to act as a communications relay station.

Being hardware-agnostic, the system can simultaneously handle AUVs, UASs and manned or autonomous land vehicles and airborne surveillance aircraft. With its open architecture, Lattice is designed to integrate diverse platforms and payloads made by separate industry partners for different missions, and then adapt rapidly to team as well as individual equipment configuration changes as threats and technologies evolve.

“At recent exercises, we integrated with systems like the Ocius BlueBottle and that was very easy to achieve,” says Arnott.

Lattice for Mission Control is designed for high-level operations. Putting a team of suicide drones over a target is relatively simple and uses only a small fraction of the system’s capability. Arnott likens its role more to that of the conductor of an orchestra, getting different players to do specific things at the appropriate time.

And it is challenging some of Australia’s traditional capability development and acquisition processes. Future warfare will require greater flexibility and the ability to re-task and re-configure networks and systems of systems, and therefore their managing or controlling C3I systems, quickly. This is an area of opportunity for Lattice for Mission Control, believes Anduril.

Arguably, the entire ADF C3I system, as it adapts to AI, autonomy and blending autonomous and manned systems, is going to need some form of flexible, rapidly configurable C2 system for its sensors and effectors that isn’t based on rigid, proprietary standards. The quick and low-risk way to do this may be to acquire something like a flexible, software-based C2 system off the shelf, something that is attractive to non-traditional vendors as well as to the traditional prime contractors.

“Next generation command and control – Agile C2 as it’s been called – is exactly what we’re building this thing [Lattice for Mission Autonomy] for,” says Shane Arnott.

Warfare is going to be much more dynamic, he believes, so the ability to shrink a C2 system into a small, portable package will be important. This is possible thanks to the ability to put a C2 centre on a laptop computer, to quickly establish secure communications links and to be able to share command and control responsibilities dynamically.

Implementing a new capability development and acquisition approach to exploit this flexibility and rapidity of fielding requires a different mind-set among customer groups and operators. And the DSR seems to have given the disruptors in Defence the license to think in these terms.

The DSR talks about fielding a Minimum Viable Capability (MVC), notes Arnott: that’s Silicon Valley-type thinking, which he welcomes warmly, and he cites Ghost Shark as a good example.

This is a complete contrast to Defence’s current elaborate capability development and acquisition process, and the disruption goes beyond simply fielding an MVC.  At present, to achieve flexibility in C3I usually means integrating lots of different systems acquired at different times from different vendors – and then re-integrating replacements for at least some of them when something becomes unsupportable, or technology growth or an emerging threat renders an existing system useless.

Very heavy lock-in, exquisite connectors and data links and other proprietary things make life much harder for integrators: they are not good for the war fighter and they’re not good for innovation, says Arnott.

So, he welcomes the announcement earlier this month of Defence’s Advanced Strategic Capability Accelerator, ASCA, whose job is to foster innovation by Defence, industry and academia and then field the fruits of it rapidly.

“We’re definitely big supporters of ASCA. We think it’s great that [Chief Defence Scientist] Professor Monro will have oversight on all things innovation. That’s a really smart move for the nation.

Importantly, he says, ASCA’s governance board, which includes Professor Tanya Monro, the Deputy Secretary heading Defence’s Capability Acquisition and Sustainment Group, Chris Deeble, and the Vice Chief of the Defence Force, will ensure that resources are invested in programs for which there is a genuine need and for which a mechanism to pull them through into operational service exists.


Although Anduril Australia is focussed on development of Ghost Shark, the company is also looking at potential markets for its Air-Launched, Tactically Integrated Unmanned System (ALTIUS) multi-mission loitering munition. Versions of ALTIUS are reported to have seen service in Ukraine, but Anduril won’t comment.

The ALTIUS 600 Autonomous Air System (AAS) which is reportedly in use in Ukraine and is equipped with the same Lattice OS as Ghost Shark. Image: Anduril

Created originally by Atlanta-based Area-1 Inc, which is now an Anduril subsidiary, the 12.3kg tube-launched ALTIUS features fold-out wings, empennage and propeller and can carry an ISR, EW, communications relay, SIGINT payload as well as a warhead. It can be launched from a helicopter or another drone, or from a ship, land vehicle or static site. The baseline ALTIUS-600 model has a range of up to 440km and an endurance of more than four hours. The much larger ALTIUS-700 has a range of up to 500km, a reduced endurance of more than two hours and a maximum weight of 30kg, bringing it more into the multi-domain operational environment.

It is equipped with the Lattice OS and Lattice for Mission Autonomy and is manufactured at scale so that the price point makes a sacrificial smart drone economical as a tactical option.

Commercial manufacturing approaches make it possible to achieve scale and affordability because, unlike traditional defence procurement, you’re not buying small numbers of exotic and expensive things, says Arnott. “You’re not getting any scale benefits in your supply chain when you’re doing that, versus these things which are built in their tens, hundreds, even millions – you’re getting much more of the benefit we see in our iPhones and laptops.”

The company has opened a new, expanded manufacturing facility in Atlanta that will be the development and scale manufacturing hub for ALTIUS as well as its other Autonomous Air Systems (AAS). Measuring 180,000ft2, the new facility is expected to accommodate 300 personnel, rising to 500 in due course to meet current and anticipated demand.

Evergreening and the cycle of renewal

What is the design life of a system such as ALTIUS? What is the cycle of renewal? Shane Arnott talks about the RAN’s own concept of ‘evergreening’ which reflects the fact that the threat is moving so fast that a rapid technology refreshment cycle has become necessary.

“Some of our applications, we can update software within 24 hours, depending on the security environment – there are very few companies that can do that,” he says.

Looking at hardware as part of the ‘evergreening’ process, Anduril expects to reinvent the whole system from time to time: “We could see a completely new set of hardware every two years. We’re able to do that affordably, the customer is able to accept it into service incrementally and then that frees you up to accept new technology.”

“It’s going to be an active fight in the future,” warns Arnott. “There’s going to be replenishment, there’s going to be upgrades: you won’t be operating things for ten years plus.”

The trick is to maintain computer and hardware consistency so that whenever something new comes along, Anduril can still drop its OS in to it and it blends seamlessly into the sensor and effector fleet.

As soon as Australian systems like Ghost Bat and Ghost Shark enter service this cycle will ramp up. Anduril is looking at rapid refreshment: the software will be refreshed every few days or weeks, hardware will be refreshed every few months or years; new payloads will be inserted as you go.

“Also, there’s going to be attrition,” he reminds us: “Gone are the days when you say, “Oh we need 72 JSFs, or whatever, and that’s all we’re going to need.” By their very nature, these things will be lost in action. So what is your strategy for renewal, for the supply chain? What does that look like? What is your ability to surge as a result?”

Anduril aims to deliver answers to these questions by disrupting traditional capability development and acquisition models and delivering useable capability quickly.

Back To Top