Abstracts and papers
Access all the papers presented at ICSILP 2018. Papers are listed in the order they appeared in the program.
Autonomy Enhancing Land Power
Chair: Robert Bolia, DST – 5 September PM
AI in weapons: The moral imperative for minimally-just autonomy
Jason Scholz (Trusted Autonomous Systems Defence CRC, Australia) and Jai Galliott (University of New South Wales @ ADFA, Australia)
For land power to be lawful and morally just, future autonomous systems must not commit humanitarian errors or acts of fratricide. To achieve this, we distinguish a novel preventative form of minimally-just autonomy using artificial intelligence (MinAI) to avert attacks on protected symbols, protected sites and signals of surrender. MinAI compares favourably with respect to maximally-just forms proposed to date. We examine how fears of speculative artificial general intelligence has distracted resources from making current weapons more compliant with international humanitarian law, particularly Additional Protocol 1 of the Geneva Convention and its Article 36. Critics of our approach may argue that machine learning can be fooled, that combatants can commit perfidy to protect themselves, and so on. We confront this issue, including recent research on the subversion of AI, and conclude that the moral imperative for MinAI in weapons remains undiminished.
Ravos: Exercising contextually aware distributed autonomic control in land vehicles
Benjamin Campbell, Glen Pearce, Ant Perry, Brendan Sims, Mohammad Zamani, Lance Newby and Robert Hunjet (DST, Australia)
This paper explores the need for contextually aware distributed autonomic control of land vehicle mission systems. It proposes Ravos, a distributed, autonomic land vehicle mission system controller. It describes exemplar applications enabled by contextually aware autonomic control and the architectural requirements to achieve them in an integrated manner. We explain how the capabilities provided by these systems would be beneficial to land forces and conclude with a description of the planned future work required to develop Ravos.
Concept development of autonomous multi-purpose MHE for logistics and combat service support
Alex Schutz and Guy Edward Gallasch (DST, Australia)
Contemporary military logistics operations are commonly reliant on the use of material handling equipment (MHE). The large variety of MHE required to keep a base operational can be expensive, slow, and dangerous. A conceptual design is proposed for a networked fleet of autonomous material handling equipment to replace or augment existing logistics equipment. The proposed concept is capable of standardising existing material handling functionalities into a cohesive modular system. This system has advantages of ease of use, increased space optimisation, higher utilisation and reduced personnel-hours over traditional material handling practices. The feasibility of the concept is evaluated and predicted to be realisable within the next 10 years.
Quantifying and predicting human performance for effective human-autonomy teaming
Anna Ma-Wyatt (University of Adelaide, Australia), Justin Fidock (DST, Australia) and Hussein A. Abbass (University of New South Wales @ ADFA, Australia)
The challenge of effective human-autonomy teaming lies not only in development of technology to facilitate autonomy but in understanding how human capabilities can be integrated with autonomous systems. In this paper, we describe a program of work to address this challenge by taking a multidisciplinary approach to quantifying human performance and developing adaptive interfaces and intelligent control algorithms that will enable effective human-autonomy teaming.
The role of rejection within the trust calibration process: Insights from a mixed-methods human-robot teaming experiment
Tony Nguyen, Graeme Ditchburn (Murdoch University, Australia) and Justin Fidock (DST, Australia)
An important influence on the appropriate exploitation of robotic and autonomous systems (RAS) will be the human-RAS trust calibration process in the context of transitioning between different modes of control. Understanding the dynamic interaction of factors that influence calibration in such use contexts is an important research question. This paper sheds light on these constructs through a simulated experiment in which participants (n = 11) alternated between teleoperation and supervisory control whilst monitoring the robot’s performance and locating objects concurrently. Applying an integrative mixed-methods approach involving semi-structured interview questions, behavioural observations, and quantitative data from a parallel study, a narrative account of the trust calibration process and the primacy of rejection relating to trust calibration was extracted. Implications for the design of robotic systems and the training of human-robot teams are discussed.
Technology convergence innovation
Chair: Axel Bender, DST – 6 September PM
Enhancing Army training and education with the Bright Fox system
Philip Temby (DST, Australia) and Luke Thiele (Rheinmetall Electronics Australia)
The Australian Army has identified a need to develop a cognitive edge over adversaries to maintain a competitive advantage. As part of Army’s current modernisation efforts, it is seeking to enhance current training and education methods using new and innovative technologies. In particular there is a focus on enhancing individual decision making by teaching officers ‘how’ to think, not ‘what’ to think using digitised training systems. DST has partnered with Rheinmetall to develop a computer-based program called Bright Fox (BFOX) that attempts to analyse how individual commanders solve tactical problems. Using advanced qualitative data analytic methods, BFOX automatically generates a time-series trace (or cognitive profile) of an individual’s thinking to allow a detailed and objective assessment of how they generated a tactical solution. In this paper we outline the BFOX concept, describe its current functionality and initial results from user testing with Army personnel. Recommendations are outlined for future research and development to help realise BFOX’s potential as a training tool and assist Army to maintain a cognitive edge.
Biomarker-enhanced free-roam VR technology for tactical training
Eugene Nalivaiko, Frederick R. Walker (University of Newcastle, Australia), Daniel Harrison (Army Headquarters, Australia), Anthony Swinsburg (Zero Latency, Australia) and Eugene Aidman (DST, Australia)
This paper describes the first trial of a free-roam virtual reality (FRVR) technology for military training. Eight infantry soldiers performed a series of VR-simulated room clearance tasks modelled on their live tactical training. The participants performed the tasks in pairs and could see each other’s avatars and interact in VR. The complexity/stressfulness of the task was manipulated by altering the number of civilian and opposing-force avatars in the scenario and by triggering explosions of improvised explosive devices. Immersion in combat-based FRVR scenarios resulted in robust and reproducible increases in heart and respiratory rates, compared to both resting baseline and non-tactical walk-through conditions. These results indicate that FRVR technology is capable of provoking significant stress response comparable with existing combat training modalities. Repeated exposure to the same FRVR scenarios over five days reduced the cardiac and respiratory responses, thus indicating the utility of FRVR for exposure-based combat stress inoculation training, with biometrics enhancing its capacity to track training gains and measure its overall effectiveness.
Body armour with power storage capabilities
Jie Ding, Tim Bussell (DST, Australia), Caiyun Wang, Kewei Shu and Yu Ge (University of Wollongong, Australia)
Adding shear thickening fluid (STF) into ballistic fabrics to create “liquid body armour” to improve kinetic impact resistance has been proven. Here, for the first time, we have tried to turn “liquid body armour” into a battery system which is not only able to supply power, but also to provide protection against bullet impacts. Such multifunctional energy storage systems can share space and weight with existing body armour. Batteries with different combinations of Kevlar-based electrodes, Kevlar separator and shear thickening electrolytes have been assembled and their electrochemical performance was investigated. They demonstrated reasonable charge/discharge capacities. The concept of a kinetic impact resistant battery, combination of Kevlar electrodes with Kevlar separator, or Kevlar separator with STF electrolyte, has been demonstrated. However, some technical issues such as the moisture retained inside the Kevlar electrodes need to be addressed in order to improve the performance.
Development of tracked combat hybrid-electric vehicle
Hidetaka Taira, Takeshi Yoshikawa and Koki Jumonji (Acquisition, Technology and Logistics Agency, Japan)
Electric power needs are significantly increasing and the vehicular power management system becomes more and more important for combat vehicles. Hybrid electric vehicle technologies are advancing in civilian sectors and military ground vehicles obtain following advantages by implementing hybrid electric drive systems: improvement of acceleration and fuel economy, silent mobility and silent watch capabilities and enhancement of on-board electric power generation. However, to the best of authors’ knowledge, there has been no deployed hybrid military vehicles so far and the availability of hardware related papers is limited in spite of these advantages. This paper shows a system design and details of mobility testing of a Tracked Combat Hybrid Electric Vehicle (T-HEV) for the first time. Basic mobility performances for ground combat vehicles such as acceleration, top speed, gradeability and pivot turn capabilities are evaluated and advantages of a hybrid electric system are discussed. Finally, it was revealed that a series hybrid electrical drive system was applicable to combat type vehicles.
Advanced munitions: 3D printed firepower
Bryce Dolman, Andrew Hart, Iain Johnston and Chad Prior (DST, Australia)
3D printing or additive manufacture is increasingly recognised as a technology of strategic importance to Defence organisations. In the weapons context, much of the focus to date has concentrated on the production of inert componentry. However, we foresee that 3D printing of energetic materials - such as propellants, pyrotechnics, and explosives - will also provide significant weapon performance, logistics, industrial and strategic advantages in the future. The Defence Science and Technology Group has initiated a Transformative Energetics research program to advance these goals.
This paper describes research underway at the Defence Science and Technology Group to develop 3D printing manufacturing processes for energetic materials, and to design and optimise new gun propelling charges which fully exploit the flexibility of this manufacturing method.
Decision support system architecture for threat object detection in Defence environments
Canicious Abeynayake (DST, Australia)
Detection of threat objects such as Improvised Explosive Devices (IEDs) has become a serious challenge in defence environments. Currently a number of individual technology sensor systems are used to detect different aspects of these threat objects. In complex scenarios, human operators are challenged to interpret multiple sensor outputs in real time. This includes making decisions about the nature of the threat encountered and taking appropriate action. This paper presents a novel decision support framework to assist human operators in making decisions through fusion of outputs from multiple sensors with the situational awareness information available at the time.
Monitoring the age of vehicle shock absorbers
Carl Q. Howard, Nataliia Sergiienko (University of Adelaide, Australia) and Guy Gallasch (DST, Australia)
Shock absorber performance is critical to the operation of military vehicles. The Australian Army is interested in implementing health and usage monitoring systems for improved maintenance and fleet management. Shock absorbers typically break down by the failure of the seal where the rod shaft enters the main body, or the seal at the end of the piston within the body. There are few practical options for monitoring the condition of shock absorbers due to harsh operating environments (temperature, dirt, shock loading and continual vibration). Instead of monitoring the change in dynamic performance of a suspension system, it is proposed that the age of a shock absorber can be estimated by measuring the cumulative work done using a calorimetry method involving temperature sensors. This paper describes a simplified thermo-mechanical model that can be used to estimate the cumulative work done by a shock absorber, which is indicative of its age.
Upping the IQ of Army’s digital communications
Greg Judd, Rachna Lorke, Peter Boyd, Vanja Radenovic (DST, Australia) and Kevin Chan (Army Research Laboratory, United States of America)
Since the advent of digital messaging over tactical radio networks the volume and rate of information has become too much for soldiers and commanders to effectively prioritise and control without automated help. Current tactical automated information systems however, do not manage this information anywhere near as intelligently as a human would if they could. This paper describes how Defence Science and Technology Group and its international, academic and commercial collaborators are investigating how the ‘intelligence’ of these automated information management systems can be improved. Using emerging machine learning techniques, this collaborative multi-year SMARTNet research program will address the three key research challenges preventing more intelligent solutions from being fielded. After briefly reporting on lessons learnt from an initial ‘proof of concept’ experiment, the paper concludes by warning that, without a more ‘intelligent’ autonomous information management system, new game changing technologies may not work effectively in a complex and contested, peer to peer, battlespace.
Developing a ground truth high resolution hyperspectral imagery database for target and natural background EM signatures characterisation and assessment
Bin Lee, Jay Yu and Tim Bussell (DST, Australia)
Modern camouflage techniques not only require an understanding of the asset (target) material, but also the sensor and platform used to detect a target. This paper discusses the need for a new approach to the design of camouflage materials and counter-surveillance measures and the tiered surveillance platform-sensor systems that are driving this need and describes the basic concepts employed within the ‘Lelantos’ program to deliver this new approach. Examples of field data collected as part of collaborative international programs are presented and described in terms of ‘spectral fingerprinting’ of targets in an operational environment.
Future technology-concept exploration and weapon autonomy
Jaci M Pratt and Marcus J Tregenza (DST, Australia)
The need for innovation and technology insertion within warfighting is greater than ever; however, the impact of technology too often drops short of expectations due to limited utilization or operational failure. This paper discusses a program that aims to improve technology insertion through the application of collaborative operations analysis in Future Technology-Concept Exploration (FTCE). An incremental-spiral analysis methodology is being developed and applied to understand the impact, effectiveness and utility of technologies and their associated concepts of employment. The paper will discuss the concepts behind FTCE, the incremental-spiral analysis methodology, and the first program activity focused on autonomous weapons exploration.
Exploring the impact of augmented night vision systems on cognitive workload and situational awareness for dismounted soldiers
Thomas Price (Tactix Group, Australia) and Larissa Cahill (DST, Australia)
Augmented Reality (AR) helmet-mounted displays (HMD) provide the opportunity to improve task performance but also pose a cognitive burden risk. Several studies were conducted to assess the impact of number and type of HMD iconography on dismounted soldier performance. Study 1 was a pilot study involving field and simulation based assessments to compare cognitive workload for participants with and without an AR system. Studies 2 and 3 manipulated the simulation environment with number and type of icons.
Active protection system soft-kill using Q-learning
Arvind Rajagopalan (DST, Australia)
In this article, an active protection system design using a Q-learning based reinforcement learning (RL) is described and evaluated. It is applicable in the context of defending against fully observable threats that possess high mobility. The Q-learner is implemented with an Artificial Neural Network, which is a model-free RL method to solve what is effectively an optimal control problem that can be difficult to solve analytically. Numerical experiments are provided to illustrate the performance of Q-learning under various approaching directions of the high mobility threat. Finally, discussions for further improvements with applying Q-learning as well using substituting with policy-optimization RL techniques are provided.
Risk reduction in Defence range safety through digitisation exploitation
Joe Romeo (QinetiQ Australia)
The conduct of complex live firing activities is a critical enabler to training systems underpinning force generation of warfighting capabilities in preparation for operational employment. Ensuring the safety of our warfighters is paramount. The majority of land based range safety practices and procedures are managed by Army on behalf of all of the Services. The safe conduct of range activities is highly reliant on the application of a comprehensive library of doctrine for these practices. This paper discusses a clear requirement for an innovative system to enable improved efficiency and effectiveness of range safety through a validated and controlled process. The system should have controls in place, based on verified doctrinal procedures, to identify and restrict unacceptable levels of risk being planned into practices to assure the safe conduct of live firing training activities.
An experimental platform for heterogeneous multi-vehicle missions
Eric Schoof, Chris Manzie, Iman Shames, Airlie Chapman and Denny Oetomo (University of Melbourne, Australia)
This paper documents the development of an experimental platform that can be used to test and validate collaborative multi-vehicle algorithms and to explore the collective behaviors that arise. Unlike existing platforms, sensing and communication are explicitly considered, enabling realistic scenarios to be tested and prototyping of systems to be facilitated, thereby accelerating the development of next generation systems for defense and civilian applications.
Swarming logistics for tactical last-mile delivery
Sam Thornton and Guy Edward Gallasch (DST, Australia)
Last-mile delivery in a military context can often be dangerous, putting personnel and the supplies they carry at risk. The emergence of aerial ‘delivery drones’ from the commercial delivery sector highlights the possibilities of uncrewed vehicles being used in last-mile delivery. However, demonstrations of such technology have been limited to single vehicle deliveries, where only small portions of supplies can be delivered at once. This paper explores the concept of low-cost, uncrewed vehicle swarming for tactical last-mile delivery in a deployed setting. The benefits of uncrewed swarming systems over conventional methods of resupply are discussed, as well as the vulnerabilities and challenges faced by such systems.
Usability of a gesture control device for robotic platform operation
Geoffrey White (DST, Australia) and Susan Hill (Army Research Laboratory, United States of America)
This paper describes a study which examined the usability of a commercial-off-the-shelf gesture recognition device, to control a small, Parrot® robotic platform. Results indicate that participants were able to easily learn and use the gesture device to control the robot for a simple movement task, although some usability issues were identified. The limitations discovered included the need for a neutral position between gestures, the lack of variable controls, the desire for more precision of control and the impact of fatigue on the participants’ performance. These initial findings will inform future concept development and experimental research for human interaction with automated or autonomous technology.
Active multispectral camouflage panels
Kamil Zuber, Simon Firth, Peter Murphy (University of South Australia, Australia) and Vivienne Wheaton (DST, Australia)
Not being seen ‘in the field’ or in other words, a combat situation, can provide a critical advantage towards survivability and protecting military assets. Current camouflage technologies and approaches are fixed and unchangeable post-manufacture, and even when designed for terrain-specific environments, remain limited in their desired effect. We have developed active, multispectral camouflage panels based on conducting polymers that provide for visual and Near Infrared reflectivity to adapt (in real-time) to the ever-changing natural environment. This will furnish the material basis that is fundamental to an automatically adapting, bespoke camouflage solution for protecting military vehicles and assets.