Abstract
There is no
doubt that autonomous armed combat platforms will be integrated into the ground
forces of the world armies. The problem becomes how to identify friendly forces
that will work in conjunction with the armed units to act as a force multiplier
and to prevent identification counterfeiting by the opposing forces. This paper
investigates the possibility of using a system that will allow autonomous
robotic vehicles to function alongside ground forces and to supplement them. A
multifaceted approach is investigated, including the use of different
techniques and technologies to ensure correct identification and prevent
infiltration. The use of current technologies is possible, but it must also not
add any significant weight to a soldier’s load or require the addition of
external battery packs that could fail to cause a false positive for enemy
forces. The limitations are presented, particularly the range of effectivity
and how the escalation of responses will be handled by the units. Finally,
implementation of the equipment is presented including what available
technologies that may work with the proposed system.
Introduction
The armed forces of the world are engaged in a technology race reminiscent of the cold war. This time, instead of building a nuclear arsenal or large air and naval fleets, they are focused on enhancing their capabilities through the use of autonomous unmanned systems. Much research has already been conducted with air and sea units, particularly how to operate cooperatively with manned platforms (Horowitz, 2019). Recently, governments are increasing the amount of research into cooperative operations between ground troops and armored platforms (Jenks, 2017). This is the next logical step in the development of these platforms, to act a force multiplier for the military.
The problem to overcome is how to identify friendly forces that operate in the vicinity of the autonomous armed combat vehicles and to prevent them from inadvertently classifying them as the enemy. For this, the implementation of a concept that has been used in military aviation for decades, identification friend or foe or IFF, has been used by air forces across the world to identify friendly forces and to prevent targeting them with what has become the standard with over the horizon air combat and long-range standoff distances. This technology can be miniaturized and adapted to be used by ground forces that must operate on the same area of, and in conjunction with armed autonomous combat vehicles. This technology will be comprised of a multi-factor identification method including visual imagery, hyperspectral cameras, and wireless methods. No single method by itself will identify the friendly forces, but instead, a combination of the indicators are going to be required to ensure a high enough confidence level by the equipment and the ground forces that have to operate with the units. The multifaceted approach to identification will also prevent counterfeiting the signals to prevent interference with its primary mission and reduce the chance of infiltration of the ranks by enemy forces during combat.
Perceived Need
The next logical step in the development of autonomous armed systems will be on the ground, air and seaborne systems already exist (Altmann, 2017). The problem is, how to identify friendly forces that will operate cooperatively with the armed units. Unlike vehicles, humans will be harder to identify on the field, so therefore a method to positively identify them to the units will be required. And because of the nature of ground-based combat, there is the chance of infiltration within the ranks by the enemy, so this system must have more than one method to identify the soldier as a friend or foe. The equipment required for this system must also be minimal for the soldier to not encumber them with any extraneous equipment.
For this reason, the system will have multiple methods of identification, including vision-based sensors that see into both visible and non-visible wavelengths, and radio signal verification. The systems range will be limited but individual units can be linked together to form a cooperative network and share information to supplement the mapping of the theater of operation (Demim, 2017). The equipment used to identify the soldiers will have to be passive as to not interfere with their equipment or other duties. The wireless and non-visible light identification methods will most likely have a small battery that uses off the shelf technology for ease of use.
The use of simple and robust technology is required due to the conditions that the units will most likely be operating under. This also reduces the skills required to operate such a system and reduces the training required so that soldiers can concentrate on the task at hand rather than dividing their attention between identification by the equipment and performing their assigned duties.
Overview
No single method of identification will be sufficient during ground operations to ensure the safety of the soldiers. This is the reason for multiple methods of identification and to prevent friendly fire incidents. Verification will be accomplished through the use of visual cues and radio signals. Visual cues, such as a quick response code, also known as a QR code, will be picked up by a camera that can detect both the visible spectrum and the higher wavelengths such as a hyperspectral camera as the use of nonstandard wavelengths above the infrared spectrum may be used to prevent spoofing by enemy forces. These visual cues will have to be incorporated into the uniform of the soldier and the other equipment and vehicles will also have to be identified similarly. These are not new technologies per se but using them in this manner is an adaption of existing equipment to fill a need that will eventually be deployed in the battlefield.
The use of digital camouflage will allow these QR codes to be incorporated into the uniform without affecting the overall pattern or effectiveness of the clothing. These patterns could also be painted onto the vehicles (Unites States of America Patent No. 20140103123, 2013). The amount of information incorporated into them could be encoded so that if the cue is eventually discovered by enemy forces, it will not be able to be deciphered into any meaningful information. This fixed method of identification is just the first part of the equation, it must also include the other methods of recognition to prevent being fired upon.
Besides the visual cues a light beacon, that produces a flash outside of the visual spectrum, will be used. This beacon will flash at a rate that can be deciphered by the logic of the unit to assist in positively identifying the friendly force. The light will also be subdued to prevent the overload of the visual sensors. The flash rate can be set by the unit to add randomness to the pattern selection and to prevent copying by enemy forces if detected. By allowing the unit to program the flashing sequence of the beacon, it adds a layer of security that cannot be duplicated, this pattern can then be shared throughout the interconnected units in case the ground forces leave the vicinity of the originating unit. It would most likely be beneficial to have a unit designated as a master that sets the flash rates and then shares it over the network. This unit could act as a central hub for the dissemination of the security protocols across the cooperative network. The beacon could also act as an identifier for wounded soldiers or if they incapacitated, once the logic detects they are wounded, it could change the pattern of the flashing to assist medical personnel to their location to decrease the response time.
The final method of identification will involve a wireless method. This will be based on a long-range radio frequency identification or RFID method. Typically, RFID technology has a limited range, but recently there have been advances to increase the distances at which it can be read and the information that it contains (Fescioglu-unver, 2015). This information will, of course, have to be encrypted to prevent interception and decoding by the enemy. The RFID units will contain information concerning the individual soldier so that identification can be accomplished in conjunction with the other equipment that provides the visual cues. This technology will also assist the medical corps in retrieving an injured soldier, coupled with the change in color and or flashing frequency of the identification beacon, the exact location of the victim can be determined by the armored units and assist the direction of the medical corps personnel to recover them.
The use of both visual and radio cues combined ensure that the units can accurately identify the soldiers working near the units. The linking together of the units will also expand the coverage and prevent infiltration by enemy units along a unified front. Sharing this information also increases the likelihood of rapid response among a changing battlefield. Much like maps, which can be cooperatively shared among the units even if they have not been to that area before, so can the identification information.
With the advance in electronics and processing power, this system could be deployed to an autonomous armed unit with minimal equipment loading. Algorithms could be updated that are already used to identify targets visually. The RFID equipment could be added as an additional package and the data fed into the central processor. These sensor packages could essentially be plug and play once the logistics of integration are completed. For this to work effectively, the units it would be installed must have a standardized system to give it the greatest utility and adaptability. The equipment must also be ruggedized due to the harsh environment in which the units will most likely operate. There is one candidate currently that could be linked into a Beowulf cluster as it can build a neural net as previous research has verified this. Although only one proposed solution, the Nvidia Jetson AGX Xavier has this capability and was specifically designed for artificial intelligence applications (Mittal, 2019). As more autonomous systems are developed, the available hardware choices will increase and reduce the need for custom boards.
Limitations
The proposed system will not function over great distances, it is limited by the range of the RFID equipment which is approximately 2000 feet in the 433 MHz band, and is meant to prevent friendly fire incidents with autonomous armed equipment (Chruszczyk, 2016). Uniforms must also be made to accommodate the visual ques in wavelengths that the hyperspectral cameras can detect while not being evident to the opposing forces. Units that do not have the required equipment could be erroneously identified as the enemy and dispatched rather rapidly due to the quickness of the control logic.
The goal of this project is to use readily available technologies without burdening the soldier with extra equipment. The light beacons are small and programmable, and the RFID tags will require a battery that responds to the systems ping request. This battery must be robust and have sufficient life to prevent power loss and not burden the soldier with extra equipment. The problem is adapting current technologies with military equipment, that is rather durable, but most likely two or more generations behind current technology. Furthermore, this technology must also be stable to allow the longest lifespan as it will most likely be in use for years for governments to adopt the systems and then spend the capital to implement it. The system must also not be deployed as a stand-alone security system, the temptation will be there, but incorrect enemy identification could be possible and it may attack friendly forces without provocation (Horowitz, 2019). So, the system is currently limited because of the identification methods that will be required to ensure safe operation with human forces.
In times of battle that requires rapid calculations, if there is no network connection and sharing of resources between the units, information overload could occur, and response times could increase significantly. The machines will be limited by the computing power that can be installed on the units, even with advances in mobile processors, visual identification requires significant resources (Susteric, 2016). This system may be more suitable to fixed locations such as garrisons where soldiers are usually behind a fixed perimeter. This way a geofence could be established by the system and anything crossing the line that is not positively identified could be tagged as a threat by the system.
Another limitation of the system is operation in urban environments. Buildings and other structures can block the signals required for identification and reduce the effective range of the system causing a bunching up of soldiers around the equipment for added protection, this makes that formation a desirable target for the enemy. The proposed system would function better on open terrain or around remote bases that would require this type of protection.
Conclusion
The proposed system is feasible with current technology, the problem is how to implement it and the development of the algorithms that will control the systems. Of particular interest, is the embedded QR codes into the uniforms. This aspect alone would require new uniforms to be issued to the entire military, the development of the inks that will be applied so that the hyperspectral camera can pick up and read it will also require research as it must not degrade over time through washing. This may require an alternate method of using them, perhaps only placing them on vehicles and other equipment and using just the beacon and RFID tag for the ground troops.
The U.S. military has already invested heavily in the development of autonomous systems, particularly air and sea assets (Jenks, 2017). The next logical step in the advancement of these systems will be the integration of autonomous systems to fight along with ground troops and to act as a force multiplier. Examples of such systems already exist but they are purely defensive such as the Raytheon C-RAM, a developmental offshoot of the Phalanx weapon system and is used to defend fixed positions from incoming rocket and mortar fire and has demonstrated a high success rate (Soesanto, 2016).
The
development of the proposed system will be in conjunction with autonomous
offensive weapon platforms. It is not known how the military will plan to
deploy the systems they develop, such as a defensive posture to protect troop
formation flanks and so forth, or if they will be in leading positions and
acting as the head of the formation to reduce the likelihood of casualties with
front line personnel. One thing is certain that we are currently in a new arms
race to develop autonomous weapons platforms to supplement the human troops and
to act as a force multiplier. The problem is that whichever nation develops and
deploys them first will have an advantage on the battlefield in lethality and
speed at which these machines can determine friend or foe and act according to
their instructions.
References
Altmann, J. S. (2017). Autonomous Weapon Systems and Strategic Stability. Survival: Global Politics and Strategy, 117-142. doi:https://doi-org.ezproxy.libproxy.db.erau.edu/10.1080/00396338.2017.1375263
Chruszczyk, L. Z. (2016). Comparison of Indoor/Outdoor, RSSI-Based Positioning Using 433, 868 or 2400 MHz ISM Bands. International Journal of Electronics and Telecommunications; Warsaw, 395-399. doi:10.1515/eletel-2016-0054
Demim, F. N. (2017). Cooperative SLAM for multiple UGVs navigation using SVSF filter. Journal for Control, Measurement, Electronics, Computing and Communications , 119-129.
Fescioglu-unver, N. C. (2015). RFID in production and service systems: Technology, applications and issues. Information Sciences Frontiers, 1369-1380. doi:10.1007/s10796-014-9518-1
Horowitz, M. C. (2019). When speed kills: Lethal autonomous weapon systems, deterrence and stability. Journal of Strategic Studies, 764-788. doi:https://doi-org.ezproxy.libproxy.db.erau.edu/10.1080/01402390.2019.1621174
Jenks, C. (2017). THE GATHERING SWARM: THE PATH TO INCREASINGLY AUTONOMOUS WEAPONS SYSTEMS. Jurimetrics, 341-359. doi: http://ezproxy.libproxy.db.erau.edu/login?url=https://search-proquest-com.ezproxy.libproxy.db.erau.edu/docview/1965542184?accountid=27203
Mckinney JR, J. A. (2013). Unites States of America Patent No. 20140103123.
Mittal, S. (2019). A Survey on optimized implementation of deep learning models on the NVIDIA Jetson platform. Journal of Systems Architecture, 428-442. doi:https://doi-org.ezproxy.libproxy.db.erau.edu/10.1016/j.sysarc.2019.01.011
Soesanto, S. (2016). US Missile Defense in the Age of Everything: From BMDS to IAMD. Honolulu: Center for Strategic and International Studies. Retrieved from http://ezproxy.libproxy.db.erau.edu/login?url=https://search-proquest-com.ezproxy.libproxy.db.erau.edu/docview/1787154662?accountid=27203
Susteric, J. W. (2016). Towards a unified multiresolution vision model for autonomous ground robots. Robotics and Autonomous Systems, 221-232. doi:https://doi-org.ezproxy.libproxy.db.erau.edu/10.1016/j.robot.2015.09.031
Patrick Carroll 