Over the last ten to fifteen years, there has been a rapid increase in the use of digital communication networks to share data between military systems. NCW in the US [7] NEC in the UK [1] allows a commander to link many sensors spread across a number of platforms, and then to distribute the resultant picture to personnel at the front line. This sensor-to-shooter capability can provide real-time situational updates to weapon operators [9], the main aim being to enhance effectiveness whilst reducing the risk of collateral damage/casualties.
The ultimate example of the NCW approach is the use of UAVs and remotely op- erated weapon systems. This is a major area of growth for the US [134], the UK [135] and an increasing number of countries around the world. The use of such platforms raises a number of legal and ethical issues that have been the subject for much recent debate [136]. One of the chief concerns in such discussions is the role of the operator and the ability of the operator to make informed, reasoned and ethical decisions when he/she is remote from physical danger [137]. This is an important area of active research that touches on ethical/legal constraints, human-computer interaction, the psychology of decision-making and the physical limits of the technologies involved [138]. In such complex multi-faceted systems, it is crucial that great care is taken to ensure that all factors are addressed.
A key decision in the development of any remotely operated system is the level to which tasks are delegated to an automated or autonomous sub-system, with the hu- man operator retaining supervisory control over the whole system. The distinction between automated and autonomous sub-systems is important. An automated system reacts to input data in a predefined and well-regulated way, whilst an autonomous
system is able to make reasoned decisions, based upon a range of possible input data. Automated systems are common; ranging from simple central heating systems to com- plex aerospace flight control systems. Autonomous systems are rapidly becoming more common; as physical robots that can sense, explore and reason about their environment and software internet bots to take much of the drudgery out of repetitive analysis tasks.
It is essential that a human operator retains the final decision regarding the prose- cution of a valid target because only a human operator has the legal (and ethical) authority to commit a weapon to a target. Whilst current doctrine and RoE require an operator in the loop, electronic systems are required to provide all of the relevant information to enable them to make their final decision within operational, legal and ethical constraints. It is important, therefore, to ensure that the information is timely, appropriate, and that it adapts to the changing situation on the battlefield.
An evolving area of applying automation techniques to support military decision mak- ing is though automatic task allocation problems. If there are a given number of assets, with different types of effects that can be delivered, how best should they be distributed in a given scenario for optimum effect and efficiency? There are a range of different optimisation techniques that can be applied to these types of distribution problems where the aim is to either minimise or maximise some function. The scenario used for configuring the task allocation software was focussed in the air-to-surface weapon domain. A number of aircraft, with different payloads, were available to be tasked amongst a mixed set of targets that require different effects.
This chapter presents an investigation of the applicability of resource-task allocation algorithms when run in a real-time environment and incorporated into an operator ter- minal. The resource-task allocation algorithms take into account the resources available for deployment; including the physical limitations on aircraft and guided weapon sys- tems moving in to engage a target, the operational constraints around using certain types of weapon against certain targets, the occurrence of targets of opportunity and the need for sensors to be deployed to provide positive target identification and poten- tially laser designation of the target [13].
4.1.1 Remote Operation of Multiple Air Platforms
UAVs or Unmanned Weapon System (UWS)s in use today tend to be remotely piloted systems, with some automated systems largely responsible for the non-critical elements of the flight management systems. An operator might be responsible for flight, weapon and sensor systems. Some advanced reconnaissance UAVs have a wide range of sensors, including multiple wide field of view imagers, multi-spectral thermal imagers, imaging
radars and electronic warfare systems. Each of these systems often requires a dedicated operator to manage the sensor, make assessments and relay the intelligence gleaned.
Weapon loads for armed platforms tend to be more limited than their sensors. An armed UAV will have a limited number of guided weapons that can be deployed against ground targets or in self-defence, often far fewer weapons than a manned platform: a mixture of precision guided bombs, anti-armour missiles and air-to-air missiles. Added to this is a new generation of small UWS, loitering munitions, which have a limited du- ration compared to a full-sized UAV but far longer flight duration than a conventional weapon system.
This chapter examines a situation where the current generation of remotely piloted systems has advanced to a stage where more of the flight-critical tasks can be dele- gated to autonomous systems and a remote operator has more of a supervisory role. This supervisory role could cover a number of different UAVs and UWSs and may involve a range of different tasks and targets. Tasks may be subject to different opera- tional constraints (e.g. risk of collateral damage, priority tasks due to troops in contact, or laser designation required). The air platforms may have different capabilities (maxi- mum speed, duration, time on station), and each platform may have a different weapon load, which depends on the weapons loaded at launch and the targets/tasks that have been assigned previously to that platform.
As already discussed in Chapter 2, research in the field of remote weapon operations for TLAMs [36] has found that a remote operator can supervise up to 12 weapons, but that operator effectiveness is significantly degraded when an operator is required to supervise 16 weapons. Similar performance degradation has been noted for ATC staff supervising similar numbers of aircraft [101]. In both of these examples, the platforms that the operators/controllers were responsible for were of a similar type and had similar roles. For example, all weapons of the same type in the TLAM scenarios and all piloted aircraft with similar performance in the ATC example. The factors associated with the different platforms/weapons and the tasks complicate the job of the operator and are likely to lead to reduced operator effectiveness if too many platforms/tasks are assigned.
In light of these studies, the algorithms discussed, although designed to deal with an arbitrary number of platforms and tasks, are optimized to be able to provide infor- mation to an operator in real-time (at a refresh rate of approximately once per second or once per two seconds) for up to about ten platforms and a slightly larger number of tasks. It would not be possible to guarantee real-time information for arbitrary num- bers of platforms and tasks because the allocation problem is an example of optimal
scheduling, which is known to be NP-complete. The time taken to find an optimal solution therefore scales in a non-polynomial way as the number of tasks and assets increases (assuming a deterministic algorithm).