Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 17
2 Operational and Technical Requirements Historically, the acquisition of military systems has been driven by two activities: development of new technologies that apply to military missions (technology push); and definition of new battlefield requirements by operational users (requirements pull). For today’s transformational Future Combat Systems (FCS), both technology push and requirements pull are being synergized by the Army, which is considering unmanned vehicles as a key enabler of increases in force effectiveness, protection, and economy. In the case of unmanned ground systems in particular, there has been significant technology push over the past several decades. As early as World War II, unmanned platforms were experimentally evaluated for such missions as minefield breaching. By the early 1980s, robotic systems to support space exploration and unmanned aerial vehicles were also being developed. By the 1990s several programs within the Defense Advanced Research Projects Agency (DARPA) and the Unmanned Ground Vehicle (UGV) Systems Joint Program Office were developing technologies to support unmanned ground vehicles. In planning military operations the warfighter considers hardware capabilities integrated with doctrine, force structure, and training. The Training and Doctrine Command (TRADOC) centers and battle laboratories study the battlefield impact of new systems and concepts and translate user insights into operational architectures and requirements. Because many operations performed by unmanned systems are the same as those provided by existing forces, the transformational nature of these systems has often been ignored. Although a number of unmanned system concepts have been explored, there have been few requirements defined or doctrinal issues evaluated. Much needs to be done by the warfighter before a reasonable understanding of the impact of unmanned systems on both FCS and the Objective Force can be assessed. Because the committee could find no specific operational requirements for UGVs in support of FCS, it became enormously difficult to establish a logical technology evaluation. A virtually unlimited number of concepts exist for unmanned systems, ranging from mechanical insects to robotic trains, and practically all have at least some military potential. The committee reasoned that UGVs for FCS would have to be ground mobile and have other compelling capabilities to be desired as part of the FCS, and decided to define four capability classes (teleoperated, semiautonomous, platform-centric autonomous, and network-centric autonomous). It then defined a set of notional capabilities to use as compelling examples of UGV systems. These postulated examples were essential to guide the evaluation of UGV technologies and to focus technical issues. By articulating the details on specific required capabilities, the examples provided “marks on the wall” that enabled the committee to assess applicable technology readiness levels (TRLs). OPERATIONAL REQUIREMENTS The Army requirements determination system has evolved over the last several years with the increased influence of joint operations, the publishing of a Joint Vision (Joint Vision 2020) by the CJCS and establishment of the Joint Forces Command (JFCOM), responsible for developing joint concepts and recommending joint requirements. The TRADOC commander is responsible for developing and publishing the Army Capstone Concept (TRADOC Pam 525-5), which becomes the guide for all other concept development activities. Integrating and supporting concepts are developed by TRADOC centers and schools and are published in a series of TRADOC pamphlets. The Future Operational Capabilities (FOC) document published in TRADOC Pamphlet 525-66 is a statement of the required operational capability needed by the Army and is intended to help the Army Science and Technology activities as well as industry research and development initiatives. A review of this document shows few references to unmanned ground systems,
OCR for page 18
and those are mostly confined to intelligence, logistics, explosive ordnance detection, and mine clearing. The FOC document includes the “autonomous unmanned capability to achieve total situational awareness (on the ground or in the air), evaluate data received, develop courses of action consistent with the commander’s intent, and employ combat power (lethal and non-lethal “smart” munitions) to achieve the commander’s objectives. This “economy of force” element will control terrain, reduce the risk to soldiers in certain areas, and complement and maintain maneuver dominance at the strategic, operational, and tactical levels. Additionally, this capability will substantially enhance peacemaking and peacekeeping operations” (TRADOC, 1997). On November 21, 2001, a formal request for proposal was released for a lead system integrator (LSI) for FCS, which included a Draft Mission Needs Statement and a Statement of Required Capabilities (SORC). These documents have provided some clarity to the definition of capabilities required for UGVs as part of the FCS. During a two-day seminar on the Objective Force and FCS in November 2001, the Army defined threshold-level capabilities for the FCS (TRADOC, 2001a,b) to include Manned and unmanned ground, air, and space means to extend vision beyond line of sight to gain timely combat information through passive and aggressive RSTA (reconnaissance, surveillance, and target acquisition) networked into an integrated common operational picture (COP) for unprecedented situational awareness and understanding. Integrated synergistic use of joint and Army manned and unmanned, air and ground RSTA to gain and maintain contact with enemy elements and to provide high-resolution combat information on terrain and weather. Robots to perform manpower intensive, high-risk functions, such as RSTA missions in urban operations (inside buildings and the subterranean dimension) and reconnaissance/reduction of minefields. Revolutionary means of transporting and sustaining people and materiel to leverage new ground and aerial concepts for delivery, including manned and unmanned systems. Mule-like robotic capability to perform a variety of sustainment/replenishment functions on a highly agile, light but survivable platform to include: carrying dismounted soldier loads operating in terrain requiring dismounted operations performing non-standard Casualty Evacuation and other services, such as battery recharging delivering classes of supply from battalion through company to the soldier to include resupply of ammunition performing combat tasks such as reconnaissance of high-risk areas. The use of unmanned aerial vehicles in the current war in Afghanistan is a prime example of user confidence in unmanned systems. The use of UAVs in military operations has been studied and tested for over 40 years. Over the last 20 years the use of UAVs in niche roles (predominantly reconnaissance) has driven the development of systems as well as requirements. The recent successful demonstration of UAVs in a lethal role in a combat situation has given the user the confidence necessary to push future development of the technology. This same level of confidence must be developed in the ability of unmanned ground vehicles in order for a leap-ahead to occur. Assessment of FCS Operational Requirements The Army is at a critical stage in the development of the FCS. The LSI has been selected. Over the next 3 years, the designs for the threshold version of the FCS will be determined, with requisite technologies brought to the prototype demonstration level. The prototype FCS demonstrator is intended to be capable of performing all desired functional requirements described in the FCS mission needs statement (MNS). An Objective Force Task Force, reporting to senior Army leadership, has been created to accelerate the acquisition and deployment of FCS and other Objective Force systems. Through coordination and assessments the Objective Force Task Force has been tasked to expedite FCS-related efforts in the concepts, requirements, S&T, and acquisition communities. The task force is responsible for FCS design and preparation for the technology readiness decisions in FY2003. In FY2004, the task force will focus on achieving success of the FCS demonstrator phase and support a transition to SDD in FY2006. Key tasks are to develop and maintain the FCS campaign plan, synchronizing the plan with the Army transformation campaign plan and ensuring FCS integration into the Objective Force. The LSI for FCS is tasked to assist the Training and Doctrine Command in developing the requirements documentation. The LSI should work closely with the TRADOC battle laboratories to develop operational architectures based on the technology readiness of the many technologies included in FCS. Missions currently envisioned to be performed by unmanned ground vehicles include reconnaissance and surveillance, rescue, eavesdropping, and mapping. Unmanned ground vehicles for lethal missions, including both direct and indirect fires, have been discussed, but that prospect clearly gives some military officials cause for concern. However, to be truly transformational, unmanned ground vehicles need to be at the forefront of the FCS program. At the time of the study, neither the Objective Force Task Force nor the TRADOC had been successful in laying
OCR for page 19
out a viable plan for integrating evolving UGV technologies and capabilities into the FCS roadmap. In the absence of definitive requirements, the acquisition community has struggled to identify where to place the emphasis for technology development. In April 2002, as one of its first actions the LSI for FCS requested industry proposals for 43 different technologies for FCS, including three UGV systems: Soldier UGV—a small soldier-portable reconnaissance and surveillance robot Mule UGV—a 1-ton vehicle suitable for an RSTA or transport/supply mission Armed Reconnaissance Vehicle (ARV) UGV—a 6-ton vehicle to perform the RSTA mission, as well as a fire mission, carrying a turret with missile and gun systems. While the FCS program may lack system requirements, there is general agreement that autonomous mobility from point A to point B, known as “A-to-B mobility,” is a key if not the key enabler for future UGV systems. Requirements for autonomous mobility have consisted of describing speed of maneuver over differing classes of terrain. They do not include descriptions of which tactical behaviors may be required to perform mission functions in combat, including such actions as: Terrain reasoning—The ability to use information about natural terrain features (elevation, vegetation, rocks, water), manmade features (roads, buildings, bridges), obstacles (mines, barriers), and weather Military maneuver—Using terrain reasoning, mission, friendly and enemy locations to determine the best maneuver and selection of positions for stealth and to support mission package needs (e.g., hull down for direct fire, clear of overhead obstructions for indirect fire) Agility—Using rapid, significant changes in speed and direction to reduce an enemy’s ability to acquire and hit a UGV Self protection—Sensing threats (e.g., mines, weapon systems, humans) in sufficient time for the UGV to avoid them; using onboard weapons systems or command and control (C2) links to friendly weapons systems to neutralize an enemy. For UGVs to act as part of a system of systems, dynamic, extended range, redundant, and networked communications are essential. The FCS statement of required capabilities describes communications that are Highly integrated, self-organizing, ubiquitous, distributed, extendable, and capable of increased yet scalable data rates Open, multilayered with multiple paths that provide redundancy for assured communications, with voice and data routing around inoperative nodes without interruption Using platforms as integrated nodes that do not rely on stationary attended nodes. TECHNICAL REQUIREMENTS FOR UGV CAPABILITIES The category of ground-traversing vehicles without a human operator onboard covers a broad range of mission capabilities and degrees of autonomy with respect to command and tasking functions, terrain reasoning, military maneuvering, and mobility design. For this reason and to facilitate its analysis the committee characterized four generic classifications of UGV capabilities based on relevance to potential Army missions and level of autonomy and the challenge required to implement. The classes are described, with a specific example system for each, in the sections that follow. Table 2-1 lists the UGV capability classes with potential mission function applications. Each of the four classes (teleoperated, semiautonomous, platform-centric autonomous, and network-centric autonomous) varies in its need for different UGV technologies. For example, the dependence on technology in the communications area varies as follows: Teleoperated ground vehicle (TGV): high requirement at all times Semiautonomous preceder/follower (SAP/F-UGV): mostly moderate requirement (placing “breadcrumbs”), except when it moves off course or when a crisis situation (e.g., minefield, enemy attack) arises Platform-centric autonomous ground vehicle (PC-AGV): little need for human control, minimal connectivity requirements while executing its mission Network-centric autonomous ground vehicle (NC-AGV): little need for human control, high need for network connectivity. Table 2-2 summarizes the committee’s assessment of the relative dependence of relevant technology areas to each of the defined UGV classes. As can be seen, differences also exist in the other applicable technology areas including perception, navigation, planning, behaviors and skills, learning/ adaptation, human–robot interaction, mobility, power/ energy, and health maintenance. UGV CONFIGURATIONS The four capability classes categorize UGVs in order of increasingly complex military applications. For each class the committee developed an example military application to provide a “mark on the wall” against which to measure tech-
OCR for page 20
TABLE 2-1 UGV Capability Classes, Example Systems, and Potential Mission Function Applications Example System Capability Class Other Possible Applications Small robotic building and tunnel searcher (“Searcher”) Teleoperated ground vehicle Mine detection, mine clearing, engineer construction, EOD/UXO, materials handling, soldier-portable reconnaissance/surveillance Small-unit logistics mover (“Donkey”) Semiautonomous preceder/follower Supply convoy, medical evacuation, smoke laying, indirect fire, reconnaissance/surveillance, physical security Unmanned wingman ground vehicle (“Wingman”) Platform-centric autonomous ground vehicle Remote sensor, counter-sniper, counter-reconnaissance/infiltration, indirect fire, single outpost/scout, chemical/biological agent detection, battle damage assessment Autonomous hunter-killer team (“Hunter-Killer”) Network-centric autonomous ground vehicle Deep RSTA, combined arms (lethal direct fire/reconnaissance/indirect fire for small unit defense or offense), static area defense, MOUT reconnaissance EOD/UXO = explosive ordnance disposal/unexploded ordnance; RSTA = reconnaissance, surveillance, and target acquisition; MOUT = military operations in urban terrain. nology maturity levels in each of the relevant technology areas. It is emphasized that these examples are not and should not be interpreted as recommended Army operational requirements. They are intended to illustrate the interplay of the applicable UGV technologies, as well as to show when the levels of technology could be developed to achieve reasonable military capabilities. In the following sections each example application is described in terms of an overarching concept, operational approach, basic capabilities, and UGV-human interface. These descriptions will be the basis for subsequent analysis to determine when the various technologies will be sufficiently robust to support a system development (i.e., reach Technology Readiness Level 6 [TRL 6]). See definitions for Technology Readiness Levels in Table 4-1 of Chapter 4. Teleoperated Ground Vehicles In teleoperation a human operator controls a robotic vehicle from a distance. The connotation of teleoperation is that the distance is or can be great enough that the operator cannot see directly what the vehicle is doing. Therefore, the operator’s information about the vehicle’s environment and its motion in that environment depends critically on sensors that acquire information about the remote location, the display technology for allowing the operator to visualize the vehicle’s environment, and the communication link between the vehicle and the operator. The operator controls the actions of the vehicle through a control interface (Murphy, 2000). For the purposes of this report, the operator of a teleoperated ground vehicle (TGV) is assumed to be responsible for the majority of the command and tasking functions for the vehicle and its mission package. Control is similar to piloting a UAV. A TGV has no onboard terrain reasoning or military maneuvering capability, nor does it access this information from any other source. The operator conducts all cognitive processes. The sensors onboard the vehicle and the communications link allow the operator to visualize the TABLE 2-2 Relative Dependence of Technology Areas for Each UGV Class Need/Relevance Technology Area TGV SAP/F-UGV PC-AGV NC-AGV Perception For A-to-B mobility 2 3 5 4 For situation awareness 0(2a) 0(3a,b) 5 5 Navigation 3 5 5 5 Planning For path 0(2b) 3 5 5 For mission 1 1 4 5 Behaviors and skills Tactical skills 1(2b) 1(2b) 4 5 Cooperative robots 1 3 5 5 Learning/adaptation 1(2b) 3 3 3 Human–robot interaction 5 2 4 4 Mobility 5 5 5 5 Communications 5 3 3 5 Power/energy 5 5 5 5 Health maintenance 1 3 5 5 TGV = teleoperated ground vehicle, SAP/F-UGV = semiautonomous preceder/follower ground vehicle, PC-AGV = platform-centric autonomous ground vehicle, NC-AGV = network-centric autonomous ground vehicle. aNeeded during crisis. bNeeded during lack of communication with operator. Key to Ratings 0 = no need 1 = low need 2 = below average need 3 = average need 4 = above average need 5 = high need
OCR for page 21
UGVs location and movement within its environment through information displays, which typically include: Screen display(s) of the TGV’s location using a geolocation system Images of the vehicle’s environment based on data transmitted through the communications link from sensors onboard the TGV. The operator may also have direct line-of-sight observation of the TGV, either with the unaided eye or with optical devices. Missions appropriate for TGV capabilities include minefield breaching, mine and ordnance clearing, tunnel reconnaissance, and some military operations in urban terrain (MOUT). TGVs come in all sizes. Example 1: Ground Vehicle Building and Tunnel Searcher (“Searcher”) Overarching Concept. The world is becoming increasingly urbanized. The Army will find more and more situations where enemy forces are attacking U.S. forces from urban or rural buildings, tunnels, culverts, caves, and similar confined areas. Dismounted troops will have to clear areas such as these to regain control of the terrain. These close, cramped areas favor the defender and can cause extremely dangerous situations for soldiers tasked with the clearing mission. A teleoperated Ground Vehicle Building and Tunnel Searcher (“Searcher”) could be of significant assistance in accomplishing these tasks. The Searcher would be a small ground vehicle that would carry high-resolution sensors and other lightweight payloads. The soldier-operator would be equipped with visual display and control devices such as a joystick and touch pad. Communications between the soldier and the ground vehicle would be by radio or wire as the situation dictated. Operational Approach. The Searcher would be small and light so that the ground vehicle and all associated payloads could be carried by a single dismounted soldier. The self-contained power supply would be sufficient for the Searcher to climb stairs and search all hallways and rooms in a typical 10-story urban building. It would also be able to enter and search tunnel, culvert, or cave complexes (out to 1 km and back) that are capable of being traversed by a small adult human. It would be capable of automatically righting itself in the event of a rollover. The Searcher would travel at variable speeds on all surfaces, up to the speed of a running soldier on flat surfaces. The Searcher would be weather resistant, so as to be able to operate in locations exposed to the weather or in buildings in which fire-fighting sprinkler systems have been activated. Basic Capabilities. At the most basic level, the teleoperated Searcher’s every move would be controlled by an operator. The basic payload would be a package consisting of any mix of infrared (IR), visual, acoustic, or other sensors. The sensor input would be transmitted to a display held by the operator. The display would provide high-level resolution for the operator to quickly identify humans, weapons, booby traps, supplies, and obstacles. At a minimum the Searcher would have an arm and manipulator that would allow it to open unlocked doors, safely detonate booby traps, and move small objects. Additionally, the Searcher would be able to mark areas or rooms that have been searched and deemed clear at the time of the search. The Searcher would be capable of carrying non-lethal payloads that could be detonated by the operator as necessary. The Searcher, upon being directed to do so, would be able to project limited synthetic voice commands either in English or in the appropriate foreign language. If communications were lost with the operator the Searcher would go into a fail-safe mode. UGV-Human Interface. The Searcher’s “level of initiative” is that it would normally wait until it is told what to do by its operator.1 The amount of control required from the operator would be continuous. Wire or radio frequency (RF) would allow real-time communications between the operator and Searcher. The Searcher would require one trained dedicated operator during operations. The Searcher would be reliable enough to be maintained by the operator and not more than one additional technician. Table 2-3 summarizes the basic capabilities of the Searcher UGV. Semiautonomous Preceder/Follower UGVs Like the TGV, semiautonomous preceder/follower (SAP/F) UGVs can come in all shapes and sizes. They are characterized by limits on the scope of autonomous mobility. Follower UGVs are the focus of current Army development and demonstrations. Preceder UGVs are follower UGVs with advanced navigation capability to minimize the need for operator interaction to achieve A-B mobility. For the purpose of description in this report a SAP/F UGV would traverse its environment by following a trail of markers (often called “breadcrumbs”) left by a “leader,” which could be a dismounted human, a manned vehicle, or an autonomous vehicle. It would use some cognitive processes to select the best route from marker to marker. For example, the onboard processing could determine the heading to the next breadcrumb using geolocation information and simple terrain reasoning. The terrain reasoning might include identifying a road and its edges, traversable paths across open terrain, obstacles to be avoided or negotiated, 1 Covey et al., 1994 defines six levels of initiative: 1. Wait until told; 2. Ask; 3. Recommend; 4. Act and report immediately; 5. Act and report periodically; and 6. Act on own. These are used to describe each of the example systems.
OCR for page 22
TABLE 2-3 Searcher: Basic Capabilities for an Example of a Small, Teleoperated UGV Function Basic Capabilities Mobility • Day and night • Climbs stairs and searches rooms in urban buildings at least up to 10 stories • Searches tunnels, culverts, caves out to 1 km and back • Variable speed based on situation but up to speed of a running soldier on flat surfaces Mission packages • Sensors with sufficient resolution to allow an operator to quickly identify humans, obstacles, and other information while searching buildings, tunnels, or other enclosed areas • Manipulators for opening doors and moving small objects • Marking system to indicate to follow-on soldiers which areas were cleared and deemed safe at the time of search • Synthetic voice projection • Non-lethal self-protection Communications • Wire and/or RF from UGV to an operator’s hand-held display Human control • Control by joystick, touch screen, or similar type input device • Continuous for planning and navigation Automated UGV control and decision making • Climb stairs Other • Weather proof • Self-righting Human support • Maximum of one operator and one maintenance technician per Searcher RF = radio frequency; UGV = unmanned ground vehicle. and other path-planning elements. Because the follower UGV would move through a “known environment” that has been successfully traversed by its leader, the leader is assumed to possess the majority of cognitive processes and makes decisions for military maneuvering. The leader would also have some degree of local situational awareness of its following UGVs through sensing modalities similar to those available to an operator of a TGV. The sensor suite is typically more complex than that on a TGV and may include: A geolocation sensor Daytime and nighttime viewing cameras (for leader override) Laser detection and ranging (LADAR) and multispectral sensors (providing digital representations of terrain and obstacles near the UGV) Foliage-penetrating sensors (to assess trafficability through grass and other light vegetation). Missions appropriate for follower UGV capabilities include (1) a soldier’s “mule” to carry weapons, ammunition, and other items cross-country behind dismounted soldiers; (2) logistics resupply vehicles to follow a leader vehicle in road-traversing convoy mode (sometimes called close following); and (3) logistics resupply cross-country (including poor roads and paths) following a leader vehicle by an interval of minutes to hours. A semiautonomous preceder UGV represents a step up in mobility autonomy from follower capabilities. The preceder must have sufficient autonomy to move in advance of its controller, which could be a dismounted soldier, manned vehicle, or autonomous vehicle. It would have sufficient cognitive processes onboard to select the best route to traverse an objective designated by the controller without following a breadcrumb trail (traversing “new ground”). The controller could provide course correction updates to the preceder, either at the controller’s option or upon request from the UGV. However, the preceder should be able to achieve its traverse objective for its normal mission/environment envelope with few calls to the controller for help. The cognitive processes of a preceder would be focused on: (1) determining heading using general instructions from the control vehicle; (2) geolocation information about itself, other UGVs, and the controller; and (3) complex terrain reasoning (but not as complex as the controller’s capability). It would also have limited capability to support the command and tasking needs of its mission package. The controller would continue to have the majority of cognitive processes associated with the mission: friend, foe, or neutral (FFN) locations; higher-level terrain reasoning functions (such as identifying “no go” terrain); and military maneuvering. The controller would also have local situational awareness of the environment and could intervene and override movement decisions. The sensor suite of the preceder would be essentially the same as that of a follower. Missions appropriate for SAP/F-UGV capabilities could include (1) RSTA missions 1–5 km in advance of the
OCR for page 23
controller’s position; and (2) forward fires or supply prepositioning up to several kilometers in advance of the controller. A preceder could possibly lead one or more followers by dropping the breadcrumbs for its followers and perhaps maintain some geolocation awareness of them. Local situational awareness for each of the followers, with the capability for command intervention and override, would remain with the controller. Example 2: Unmanned Small Unit Logistics Mover (“Donkey”) Overarching Concept. Since the time of Caesar, it has been pointed out that the battlefield load carried by the dismounted soldier in general and by the dismounted infantryman in particular is too heavy. A heavy load can drain a soldier of energy and critically slow down the reactions of a soldier placed in a dangerous situation. Soldiers are reluctant to carry lighter loads because the fog of war and chaos of the battlefield invariably foil the best plans to effectively conduct timely small unit resupply. For want of another box of ammunition or antitank weapon or night vision goggles soldiers may have perished. This situation can be improved if there is near certitude that needed weapons, ammunition, food, clothing, equipment, and other items will be delivered in a timely manner to precisely the right place. The Donkey2 Unmanned Small Unit Logistics Mover would be a medium-sized, semiautonomous preceder/follower UGV that could lighten the soldier’s load and allow an instantaneous reaction to the fight. The Donkey would be capable of automatically following a path through urban and rural terrain from a start point to a release point. It would be able to carry a load of at least 1,000 pounds and be able to operate on a nonlinear battlefield where small troop units would constantly be moving and operating out of small, dispersed observation/fighting positions. Operational Approach. The Donkey would be capable of operating day and night under all but the most extreme weather conditions. It would be capable of crossing terrain at least as well as that negotiated by current state-of-the-art, commercial all-terrain vehicles (ATV). The Donkey would be able to carry supplies, rucksacks, and other equipment forward. Although not its primary purpose, it could carry soldiers. On return trips it could, for example, carry used food containers, items needing higher-level maintenance, batteries for recharging, and in the extreme, casualties. The round-trip distance for automatic movement along an electronic path would be at least 50 km. It would be able to automatically cross small streams up to axle level in depth. The Donkey’s speed would be adjustable based on its sensing of the terrain and/or programmed instructions for a specific electronic path. At a minimum a human would mark the initial electronic breadcrumb paths for Donkeys to follow later. On one hand the Donkey could cautiously pick its way along the electronic path through unexpected rubble or battlefield debris. On the other it could maintain speeds at least 40 km/h on electronic paths that follow roads or are in open terrain. It would be able to automatically adjust its path by up to 1 km at each problem area to get around unexpected debris or obstacles and return to the correct electronic path. The Donkey would have rudimentary sensors and range finders that could detect other vehicles or humans. Based on sensor input, it would be programmed to slow down (for safety near friendly forces) or to take predetermined avoidance actions in areas where friendly troops are not anticipated. It would have a very high probability of successfully moving from start to release point. As a resupply vehicle, the Donkey’s technological emphasis would be on simplicity, low cost, and very high maintenance reliability. In very demanding situations (e.g., during early entry operations when friendly forces are initially heavily outnumbered) one Donkey would normally be sufficient to support at least a 30-person dismounted unit. In less demanding situations (e.g., in a more mature theater when friendly forces have a significant numerical advantage) one Donkey could support at least a 200-person dismounted unit. Basic Capabilities. The basic Donkey would have a minimum requirement for higher-technology payloads. Since the probability of the Donkey’s accomplishing its mission would have to be very high, the Donkey would need a secure communications package. The package would allow the Donkey to interact with humans located near the start and release points of the electronic path. During movement the communications capability would allow operators to keep track of the exact location of each Donkey. Additionally, operators would be informed of and could take action if there was a maintenance breakdown, unexpected large area that could not be traversed, or unauthorized tampering. The communications package could allow operators to direct the Donkey, while enroute, to move from one electronic path to another or to immediately return to the start point if a change in situation or priorities occurred. To preserve bandwidth and minimize signal clutter the Donkey and base station would communicate only as needed. UGV–Human Interface. The Donkey’s level of initiative would be that it would act automatically on assigned portions of its task as it follows its electronic path about the battlefield. However, it would be capable of asking for human guidance when unprogrammed or unanticipated situations occurred. The Donkey would require some human interaction. Through use of a small keyboard, touch screen, or other input device a human would program start times, pro- 2 The name “Donkey” was selected to distinguish the example from the FCS “Mule” application.
OCR for page 24
gram paths to be followed and provide decision template information. Other human interaction would be to maintain and fuel Donkeys, as well as loading supplies on the carrying decks. As the situation dictated, Donkeys could be operated as part of a small (10–12 Donkeys) team. The number of humans on each team would be no more than one supervisor and two maintenance technicians. Table 2-4 lists basic capabilities for the Donkey. Platform-Centric Autonomous Ground Vehicles The desired endpoint of unmanned vehicle evolution has commonly been described as an autonomous vehicle. The closest dictionary definition for “autonomous” is “independent in mind or judgment,” but this definition only shows that “autonomous” is a metaphor when applied to a robotic vehicle. The implication appears to be that an autonomous vehicle can be assigned a complex task or mission and will then execute it, perhaps acquiring information from other sources as it goes but without further guidance on what to do. In developing a systems architecture for the next-generation remote-controlled vehicle, DARPA (2001) defined “autonomous” as A mode of control of a UGV wherein the UGV is self-sufficient. The UGV is given its global mission by the human, having been programmed to learn from and respond to its environment, and operates without further human intervention. The committee spent considerable time pondering where to draw an operationally useful and technologically meaningful distinction between supervised vehicles that require less and less intervention from their commander to accomplish a complex mission (the incremental evolution of supervised control) and autonomous vehicles that can receive simply expressed orders for complex missions and accomplish them without needing to be told how. A further military subtlety is the characteristic of “responsible” autonomy, since UGVs capable of lethal weapons require fail-safe interrupt or override mechanisms. For the kinds of robotic vehicle applications Army planners and developers have discussed, the committee decided TABLE 2-4 Donkey: Basic Capabilities for an Example of a Medium-Sized, Preceder/Follower UGV Function Basic Capabilities Mobility • Day and night under all but the most extreme weather conditions • Negotiates terrain as well as a current state-of-the-art ATV • Crosses water obstacles up to axle depth • Follows electronic “bread crumb” paths for a round-trip distance of at least 50 km • Variable speed but at least 40 km/h on roads or open terrain Mission packages • Cargo bed with minimum of 1,000 pounds load capacity to carry logistical supplies between a base station and dismounted troop locations • Sensors and range finders that can identify and locate other vehicles and humans Communications • Secure package allowing communications between Donkey and base station • Bandwidth and signal clutter minimized; communication occurs only when programmed, the UGV is queried, or when human guidance is needed Human control • Mark electronic paths for Donkey to subsequently follow • Load and unload cargo • Program the electronic path to be followed, start points and stop points; change paths or start/stop points while en route as situation dictates • Monitor communications from Donkey that unanticipated situation has occurred at a certain time and location; take appropriate action Automated UGV self-control and decision making • Adjusts path up to 1 km to skirt unexpected obstacles or debris • Identifies humans; slows for safety reasons when near known friendly locations; speeds away in locations where humans are identified but no friendly troops anticipated Other • Very high probability of successfully moving from start point to stop point and back • Very high levels of maintenance reliability • Emphasis on simplicity and low cost Human support • A maximum of one supervisor and two maintenance technicians will be able to operated a small (10– 12) Donkey UGV team
OCR for page 25
that two conditions, at least, must be met for a robotic vehicle to be considered autonomous. First, the vehicle must have A-to-B autonomy. A vehicle has A-to-B autonomy when it is at point A, can be given a direction to go to point B, and can get to B with no help along the way from a human operator. For Army applications the benchmark for distances between A and B is from a few kilometers to a hundred (Eicker, 2001) and should include most of the terrain conditions where the Army would operate with manned vehicles of similar mobility characteristics. The second condition that the committee set for an autonomous Army UGV is that the vehicle must be able to carry out its assigned mission in a hostile environment. As with negotiating difficult terrain, the benchmark here is that the UGV should have survivability and self-defense roughly equivalent to a similar manned vehicle sent on the same mission. For example, on a forward-scouting mission it may not need to survive a direct hit by a rocket or anti-armor round, but it should not be incapacitated by an unarmed human running up and throwing a blanket over its sensors. This self-survival capability adds another set of technological requirements. The vehicle must have local-area RSTA capabilities, beyond obstacle detection and identification for navigating, to detect the presence of potential threats and either take evasive action, stealthy maneuver, or offensive self-defense. The vehicle must be capable of identifying friends, foes, and noncombatants (or “neutrals”) (IFFN). It must have adequate lethal or non-lethal weapons for self-defense; in the event of hostile actions it should be able to survive. It must be able to refuel itself by means likely to be available, such as prepositioned or air-dropped fuel supplies or rendezvous with a fuel supply vehicle (manned or unmanned). It must have sufficient reliability and robustness to absorb and overcome the common mishaps of cross-country maneuver. It must have the higher-level cognitive processing needed to support tactical maneuvers and self-protection/self-defense behaviors. Once the committee recognized that this second requisite was essential for “autonomy” in Army applications, it became obvious that an autonomous UGV could easily have multimission capability. Relatively “dumb” mission-specific modules could be attached to the basic vehicle. Specific instructions for how to work the equipment, equivalent to a soldier’s training on a weapon or other specialized equipment, could be in software loaded into the UGV’s main computing capability when the mission package is attached. The intelligence to know when and how to employ the weapon system or equipment appropriately (software equivalent of doctrine and rules of engagement) would already be present in the behaviors that give the vehicle A-to-B autonomy in hostile environments. Example 3: Unmanned Wingman Ground Vehicle (“Wingman”) Overarching Concept. Small mechanized infantry and armor combat units normally operate and fight as teams. In mounted operations the smallest team consists of two vehicles in which a section leader (with crew) in one combat vehicle is accompanied by a subordinate wingman (with crew) in another vehicle. The section leader, following broad guidance from his superiors, usually determines routes to be followed, positions to be occupied, formations to be used during movement, and actions to be taken during various situations. In combat, when two or more enemy targets are encountered, the section leader will give fire commands to determine which enemy targets he will attack and which will be attacked by the Wingman. The Wingman is constantly observing his designated sector and reports any changes or dangers to the leader. The Wingman’s observations and communications often result in the section leader changing his initial plans based on more complete information. A Wingman could perform section missions normally assigned to a manned Wingman vehicle. The unmanned Wingman would be capable of functioning at the highest states of alert indefinitely. It would be able to routinely conduct missions that would normally be considered extremely risky for a manned system. The Wingman would be able to automatically move about the battlefield—remaining within assigned areas or on designated routes—as directed by the section leader. Through its sophisticated sensor package the Wingman would provide eyes and ears to the section leader who could give early warning of danger and increase the survival rate for the manned vehicle. The Wingman could provide 360-degree observation or focus solely on a specific sector as directed by the section leader. It would provide an RSTA capability that would allow it to recognize natural and manmade features as well as nearby people, vehicles, obstacles, and other information. The Wingman would transmit observed information to the section leader. This intelligence would allow the section leader to continue with the current plan, adjust movement, maneuver to a more advantageous position, directly attack an enemy target, request other assistance to accomplish the task, or take any number of actions as necessary. The Wingman would be able to perform continuous local security while soldiers slept or were otherwise occupied. Operational Approach. The unmanned Wingman, a medium to large platform-centric autonomous ground vehicle, would be capable of operating day and night under all weather conditions. It would be capable of moving at variable speeds, based on conditions encountered, up to at least 100 km/h on roads and in open terrain. Both the Wingman and the section
OCR for page 26
leader’s vehicle would be able to traverse urban and rural terrain and swim across ponds, lakes, or other slow moving bodies of water with equal ability. It would automatically move in relation to the section leader (precede, follow, or on a flank), as directed. The Wingman UGV would be the approximate size of the section leader’s vehicle to preclude the enemy from more easily identifying and firing at the section leader. Stand-off distances that the Wingman would achieve relative to the section leader would be based on doctrine (tactics, techniques, and procedures), local terrain, higher headquarters guidance, and the section leader’s order. The Wingman would be able to sense the section leader’s location, the location of other nearby manned or unmanned vehicles, perceive the local terrain, and automatically adjust movement direction and speed, as necessary. The section leader’s human–machine interface capability would allow him to very easily order the Wingman to move to a designated area or a designated point to conduct RSTA tasks. Depending on the situation, the Wingman would be given more or less latitude in determining routes to move to the observation position. Similarly, it could occupy and adjust its precise location in stationary positions to take best advantage of cover, concealment, and lines of sight. Additionally, it could be able to automatically tie in with manned/unmanned systems in its vicinity to achieve overlapping fields of observation and fires. It would have a continuous local security capability that would immediately signal the section leader when unanticipated movement or activities occur. The Wingman would achieve high levels of maintenance reliability. It would be able to self-diagnose and store anticipated noncritical maintenance problems for later downloading and correction. More critical maintenance or other problems that would impact on its mission would be reported immediately to the section leader. Basic Capabilities. The Wingman would be electronically tethered to the section leader through a secure local area network. The Wingman would carry a sophisticated sensor and range-finding RSTA/BDA (RSTA battle damage assessment) package. Its sensors may be any combination of seismic, acoustic, magnetic, visual, IR, or other capabilities. The Wingman would have access to a sophisticated automatic target recognition (ATR) capability that as a minimum, can distinguish between friendly and enemy combat vehicles. The Wingman’s computational capability, tied to its sensors and ATR, would allow for the rapid creation and transmission of recommended direct and indirect fire commands to the section leader. The Wingman would provide for its own survival by sensing when it is under attack from direct or indirect fire and by taking immediate programmed action, such as rapidly changing location. It would sense danger from approaching humans and be able to launch close-in non-lethal effects. UGV–Human Interface. The Wingman’s level of initiative would be to do assigned tasks and report the results of its acts. It would also alert and recommend courses of action, as programmed, for certain dangerous, complex, or ambiguous situations, but wait for guidance from a human before acting in these cases. Even though the Wingman would be capable of considerable automatic actions there would still be a need for a close human interface. The objective would be to allow the section leader to keep focused on the battle and not be distracted by the Wingman UGV. Therefore, emphasis would be on minimizing human interaction with the Wingman while maximizing the Wingman’s automatic capabilities. The assistant section leader would have the ability to teleoperate the Wingman electronically in situations where human override was necessary due to safety or highly complex situations; however, this would not be the preferred method of control and would be used infrequently. Table 2-5 lists the basic capabilities of the Wingman UGV. Network-Centric Autonomous Ground Vehicles The committee decided that a third condition not essential for vehicle autonomy was likely to be essential for autonomous UGVs to be elements of FCS. Without diminishing their A-to-B autonomy in hostile environments, the vehicles must also be competent as independent nodes in a network-centric warfare model. They must be able to receive information from the communications network and incorporate it in their mission execution. They must respond to appropriate information requests and action commands received from the network, including resolution of conflicts when requests or commands interfere with each other or with the original mission assigned to the UGV. Again the rough benchmark for operational success as an independent network-centric node is an equivalent manned system similarly tasked. Example 4: Autonomous Hunter-Killer Team (“Hunter-Killer”) Overarching Concept. Potential enemy forces, ranging from sophisticated mechanized units to Third World guerillas, need to move about the countryside, through villages, and in cities to accomplish their missions. Ambushes conducted against enemy forces can have a powerful effect. Ambushes can disrupt and slow the pace of logistics and combat operations. After experiencing initial losses to ambushes an enemy is often forced to stop movement on well-defined roads and trails that could be the most likely sites of fatal
OCR for page 27
TABLE 2-5 Wingman: Basic Capabilities for an Example of a Medium-Sized to Large Platform-Centric UGV Function Basic Capabilities Mobility • Day and night under all weather conditions • Crosses urban and rural terrain with same ability as section leader’s manned vehicle • Swims water obstacles without additional preparation • Variable speeds depending on situation but up to 100 km/h on roads or in open terrain Mission packages • Sophisticated sensors and range finders that allow Wingman to become the “eyes and ears” of its manned section leader • Sophisticated RSTA/BDA package; ATR capable of differentiating between friendly and enemy combat vehicles • Non-lethal self-protection package Communications • Secure communication package forms basis for “electronic tether” control/information sharing between Wingman UGV and human section leader • Near real-time transfer of sensor and other information to section leader Human control • Human very easily directs Wingman to new locations and describes tasks to be performed by UGV while en route and upon arrival at new location • Human monitors sensor and other input from Wingman • Actively makes go or no-go decisions on all Wingman recommended calls for direct or indirect fire • Electronically directs/overrides Wingman movements into very confined, dangerous, or complex locations Automated UGV self-control and decision making • Automatically moves in relation (precedes, follows, or on a flank) to manned vehicle as initially directed; adjusts speed and movement direction based on terrain, vegetation, nearby vehicles, or other objects • Occupies and adjusts its precise location in stationary positions; ties in observations and fields of fire with adjacent manned or unmanned systems • Automatically calls for recommended direct or indirect fire missions when sensing an enemy • Senses when under attack from direct or indirect fire and takes appropriate action • Recognizes commands to change allegiance to a different human section leader, as necessary Other • Wingman UGV is about same size as section leader’s manned vehicle • High levels of maintenance reliability • Self-diagnosis and storing of anticipated noncritical maintenance problems; immediate reporting of critical maintenance issues to section leader Human support • No more than one assistant section leader (Wingman controller) and one maintenance technician both of whom ride in the manned section leader’s vehicle ambushes. This can reduce enemy movement speed to that of dismounted soldiers. With the growing sophistication of U.S. air-delivered smart weapons and night vision capability, a thinking enemy may be forced to take advantage of urban terrain where there is better protection from U.S. observation and attack. Ambushes can be especially effective in urban areas where streets channel movement and buildings offer excellent cover and concealment for ambushing units. Additionally, an ambush can be one of the most doctrinally straightforward (albeit among the most dangerous and certainly not simple) missions conducted by a small unit. The ambush site is selected in advance on terrain that favors the friendly unit and puts the enemy at a disadvantage. Outposts can be established to provide advance warning of the enemy approach. Detailed fire plans can be made that include prearranged calls for indirect fires. Withdrawal routes and assembly areas can be planned and reconnoitered in detail. An Unmanned Autonomous Hunter-Killer Team (in an “ambush unit” scenario) can give the U.S. Army the ability to foil enemy movement on the ground, inflict heavy enemy casualties, and minimize friendly casualties in the process. The Hunter-Killer team would consist of several unmanned vehicles that would be capable of moving to an ambush site. Upon arrival they would arrange themselves in an effective observation/kill array, would be able to transmit relevant information to a human base station, and would be able to monitor the environment to know when the enemy is approaching and when he is in the kill zone. The team would kill enemy forces in the kill zone with overwhelming lethal force. Subsequently, it would take programmed actions to ensure its survival, such as moving to a new location. Operational Approach. The Hunter-Killer team would consist of at least 10 medium-sized “killer” unmanned network-
OCR for page 28
centric autonomous ground vehicles. Each killer vehicle would carry internally (in a “marsupial” manner) up to five small network-centric autonomous “hunter/observer” ground vehicles; as the situation dictated the ground hunters/ observers could be replaced or augmented with aerial hunters/observers. All UGV would be capable of operating day and night under all weather conditions. The Hunter-Killer team would be able to operate in rural and urban terrain and swim across slowly flowing rivers. Initially, the Hunter-Killer team would be programmed by humans at a base station. For intra-Hunter-Killer team communications purposes, the UGVs would be tied together through a local wireless network. Information gathered by each unmanned vehicle could be passed to others, including updating and modifying the human-provided input as long as those modifications remain within programmed hard decision rules. The ambush vehicles would have sophisticated sensors that detect humans or other vehicles and would be allowed to take evasive actions, as necessary. The Hunter-Killer team would be capable of conducting a round-trip distance between start point and ambush site up to at least 300 km. Speed of movement would be variable depending on the terrain but would be capable of reaching at least 120 km/h on roads or in open terrain. At the ambush site the medium-sized killers would launch small hunter/observer UGVs to provide area security and detailed information on approaching people and vehicles. Once the stealthy hunter/observers were in position the killers would go on power standby. The Hunter-Killer team would be able to remain on site in standby mode without human interaction for at least 30 days. When hunters/observers sensed that humans or vehicles were approaching they would automatically activate higher-power sensors and ATRs to confirm the identity of an enemy force. This information would be sent to the killers, who would move from power standby to full alert. Based on data from the hunters/observers, the killers would be able to calculate the size of the enemy force. If the enemy force were too large to ambush it would be allowed to pass, otherwise the Hunter-Killer team would risk being overpowered by the enemy and being destroyed. The killers would also determine which actions needed to be taken by an appropriately sized enemy to trigger the ambush. They would notify the human base station of actions about to happen, as programmed. Depending on the rules of engagement a human in the base station may have to signal approval. When the ambush was triggered the killers would launch an overwhelming and precisely targeted lethal response using onboard targeting systems: The general rule is “one shot, one kill.” In addition to onboard weapons, the killers would be able to automatically send fire commands to the base station or directly into the appropriate battlefield C2 networks, resulting in massive indirect fires being provided. Reactions to counterattack would be dependent on the situation but would likely initially include calling for additional indirect fire and maneuvering one or more killers to achieve a tactical and firepower advantage over the enemy. The killers would also be provided with decision criteria that would allow them to immediately move from the ambush site to a remote assembly area to await further instructions from the base station. Even if not attacked by the enemy, moving out of the ambush site quickly would be required. If time permitted, the hunters/observers would rejoin and board the killers for transportation to the next mission. Otherwise, the hunters/observers would go on standby mode or self-destruct, as programmed. Maintenance reliability would be very high. In addition to being able to self-diagnose maintenance problems the Hunter-Killer team will have rudimentary self-repair. Basic Capabilities. The unmanned Hunter-Killer team would need a sophisticated local and global terrain-sensing capability. The team’s sensors could read local terrain, vegetation, obstacles, and other information in great detail, and store this information for downloading to allow updating the global database. It would need sensors, range finders, and ATR that could accurately identify enemy forces and distinguish friend from foe from noncombatant. It would need highly sophisticated communications packages and local area networks to pass information among all necessary UGVs in the unit. It would also need to be part of a secure wide area network to communicate with its base station and other unmanned and manned systems, as necessary. It would need longer-range communications and relays to allow it to contact a distant base station or directly into C2 networks to receive human approval to take certain actions, as required, and to request additional support. Examples include passing on the arrival of a new ally on the battlefield that has similar vehicles to the enemy, the anticipation of noncombatant movements in the area, or the signing of a cease-fire. UGV–Human Interface. There would be very little human interface with the Hunter-Killer team once launched on its mission. The Hunter-Killer team would essentially be autonomous except when programmed human intervention and communications must occur. The Hunter-Killer team would also be a step beyond where each robot has self-diagnostic capabilities, because each would also do some self-repair. The hunters/observers’ marsupial UGV, for example, could actually do field repairs on the Hunters-Killers or themselves or vice versa. Another concept is that maintenance could be conducted between missions, in safe areas, by other highly specialized robots. Because of this, it is anticipated that a small team of mechanics and technicians (as few as 10) could maintain up to 5 Hunter-Killer teams. Table 2-6 lists basic capabilities for the Hunter-Killer team.
OCR for page 29
TABLE 2-6 Hunter-Killer Team: Basic Capabilities for a Small- and Medium-Sized Marsupial Network-Centric UGV Team Function Basic Capabilities Mobility • Operate day and night under all weather conditions • Operate in all rural and urban terrains • Swim across slowly flowing rivers and other bodies of water without additional preparation • Move and perform missions without active human intervention over a round-trip distance of at least 300 km • Move at variable speeds depending on the situation but up to at least 120 km/h on roads • Remain in position for at least 30 days without human intervention Mission packages • Local and global terrain, vegetation, obstacle sensing • Highly sophisticated sensors and range finders • Highly sophisticated ATR that can discriminate among vehicles (combat and commercial) and humans (friend, foe, and noncombatant) • Sensors able to read detailed terrain, vegetation, obstacle, and other data that can be downloaded upon command to update global databases • Stealth capabilities that make enemy detection of any UGV very difficult • Precisely targetable, highly lethal kill systems; “one shot, one-kill” • Lethal self-protection package Communications • Secure local area network allows all UGVs to pass information among themselves • Secure wide area network allows team to call for backup support and to communicate with base station as well as other networked systems, both manned and unmanned. Human control • Program various movement, communications, intelligence, rules of engagement, decision making, and other initial inputs • Monitor communications from UGV team for programmed reports or situations requiring human guidance • Override in case of changes in situation Automated UGV self-control and decision making • Automatic “intelligent” decision making based on programmed human instructions augmented or modified with real-time UGV sensing of the situation • Fully automated movement; capable of moving as a team or infiltrating separately • Killers able to launch hunters to gather intelligence on terrain, vegetation, obstacles, or human activity • Upon arrival at a mission location all UGVs able to close down all energy dependent systems except for the most energy efficient; capable of “waking up” other systems as the situation warrants • Only attacks enemy forces that are within its ability to devastatingly destroy; otherwise, follows programmed decision rules • Understands enemy tactics and reacts to enemy actions with coordinated UGV tactics, as necessary Other • Minimum size of one team is 10 medium-sized “killer” UGVs that each internally (in a marsupial fashion) carries up to 5 small hunter/observer ground or aerial UGVs • Very high level of maintenance reliability; self-diagnosis and repair of maintenance problems Human support • Control is by on-duty staff officer at appropriate headquarters • Maintenance beyond scope of UGV and other programming requirements performed by a small team of humans (no more that 10) to support up to five Hunter-Killer teams (approximately 50 killers and 250 hunters) Although the committee aligned the Searcher, Donkey, Wingman, and Hunter-Killer examples with the defined TGV, SAP/F, PC-AGV, and NC-AGV capability classes, respectively, many of the example applications could be performed by UGVs in more than one class, depending upon the Army operational requirements. For example, it would be possible to develop a teleoperated Donkey or a platform-centric Donkey providing more or less operational potential than the semiautonomous preceder/follower Donkey. It would also be possible to develop an autonomous Searcher with cooperative robot capabilities.
Representative terms from entire chapter: