Robotics
Publications Robotics
PublicationsRDT&E Division, Naval Command Control and Ocean Surveillance Center
NCCOSC RDTE DIV D371, San Diego, CA 92152-7383
ROBART III is an advanced demonstration platform for non-lethal security response measures, incorporating reflexive teleoperated control concepts developed on the earlier ROBART II system. The addition of threat-response capability to the detection and assessment features developed on previous systems (ROBART I and ROBART II) has been motivated by increased military interest in Law Enforcement and Operations Other Than War.
Like the MDARS robotic security system being developed at NCCOSC RDTE DIV (the Navy's Command Control and Communications center in San Diego, called NRaD for short), ROBART III will be capable of autonomously navigating in semi-structured environments such as office buildings and warehouses. Reflexive teleoperation mode employs the vehicle's extensive onboard sensor suite to prevent collisions with obstacles when the human operator assumes control and remotely drives the vehicle to investigate a situation of interest.
The non-lethal-response weapon incorporated in the ROBART III system is a pneumatically-powered dart gun capable of firing a variety of 3/16-inch-diameter projectiles, including tranquilizer darts. A Gatling-gun style rotating barrel arrangement allows six shots with minimal mechanical complexity. All six darts can be fired individually or in rapid succession, and a visible-red laser sight is provided to facilitate manual operation under joystick control using video relayed to the operator from the robot's head-mounted camera.
This paper presents a general description of the overall ROBART III system, with focus on sensor-assisted reflexive teleoperation of both navigation and weapon firing, and various issues related to non-lethal response capabilities.
Keywords: security, robot, teleoperation, non-lethal
The robotic site security (sentry) application has a number of features which match the strengths and surmount the weaknesses of conventional unmanned ground vehicles (UGVs): (1) unlike the battlefield reconnaissance, surveillance and target acquisition (RSTA) application, the operating environment is known in advance, is normally under friendly control, and can to some degree be tailored to support robotic operations; (2) experience-based costs of inventory shrinkage and non-robotic security measures provide a sound and credible basis for robotic sentry system cost/benefit tradeoffs; (3) unmanned vehicles do not get bored during long hours of surveillance, leading to reduced vigilance; and (4) unmanned vehicles don't participate in "inside jobs."
Generally regarded as the world's first autonomous security robot, ROBART I (Figure 1) was developed at the Naval Postgraduate School [1]. While rich in collision avoidance sensors, this research platform had no sense of absolute location within its indoor operating environment, and was thus strictly limited to navigating along preprogrammed patrol routes defined by the relative locations of individual rooms, periodically returning to a recharging station by homing on an optical beacon. From a security perspective, the platform could only detect suspected intruders, with no subsequent intelligent assessment capability to filter out nuisance alarms.
Figure 1.
ROBART I (1980-1982) was developed at the Naval Postgraduate School in Monterey, CA.
The second-generation follow-on to ROBART I was ROBART II (Figure 2), which also operated indoors, incorporating a multiprocessor architecture and augmented sensor suite in order to support enhanced navigation and intelligent security assessment. ROBART II was transferred to the Naval Ocean Systems Center (NOSC, now NRaD) in 1986, and used as a testbed for the development of obstacle mapping and other sensor fusion and navigation capabilities. The addition of an absolute world model allowed ROBART II to: (1) determine its location in world coordinates; (2) create a map of detected obstacles; and (3) better perform multisensor fusion on the inputs from its suite of security and environmental sensors [2]. This last feature facilitated the implementation of a sophisticated threat assessment algorithm that significantly increased the probability of detection while virtually eliminating nuisance alarms.
Figure 2.
ROBART II (1982-1992) employed 132 external sensors for navigation and intrusion
detection.
Two commercial security systems also appeared in the mid-80s. The Denning Sentry was an indoor robot for which development began in 1983, ultimately involving the investment of several million dollars and contributions from CMU robotics researchers. Fixed-path navigation using modulated near-infrared beacons, and a modified doppler microwave motion detector designed for use on a moving platform were two of the innovations pursued by Denning, but ultimately the company went out of business. (It has since been revived as Denning Branch International Robotics, but with focus on applications other than security.)
Meanwhile, Robot Defense Systems (RDS) developed the Prowler, an outdoor sentry/surveillance platform built on a commercial diesel-powered six-wheeled chassis that was capable of following a preprogrammed patrol path with limited obstacle avoidance capability. Video was relayed via microwave back to the human operator, who could override the onboard control computers when necessary. While the Prowler successfully demonstrated the ability to autonomously follow along a (slightly modified) fence line in a test for the US Army in October 1984, RDS went out of business in 1986 [3].
The Defense Nuclear Agency began a Physical Security Robotics Program in March 1983. Following an initial phase of feasibility studies, DNA has been a major sponsor of Sandia National Laboratories' mobile robotics program. A number of Sandia's robotic systems (i.e., Dixie, RAYBOT, Telemanaged Mobile Security Station (TMSS), and RETRIVR) are described by Byrne, et al., in "Military Robotics Technologies" [4].
The Mobile Detection Assessment and Response System (MDARS) is a joint Army-Navy development effort to provide an automated intrusion detection and inventory assessment capability for use in DoD warehouses and storage sites. The program is managed by the Physical Security Equipment Management Office (PSEMO) at Ft. Belvoir, VA, with overall technical direction provided by NRaD. The MDARS goal is to provide multiple mobile platforms that perform random patrols within assigned areas of warehouses and storage sites. The patrolling platforms: (1) detect anomalous conditions such as flooding or fires; (2) detect intruders; and (3) determine the status of inventoried items through the use of specialized RF transponder tags.
Separate development efforts target warehouse interiors and outdoor storage areas. The MDARS Interior Program utilizes the Cybermotion K2A Navmaster mobility base (Figure 3) equipped with additional collision avoidance, intruder assessment, and product inventory subsystems. Simultaneous control of two robots patrolling nightly within an interior warehouse environment has been demonstrated for over 12 months at a government facility in San Diego, CA. The MDARS Exterior Program, initiated in February 1993, awarded a contract for the development of the mobility platform (Figure 4) to Robotic Systems Technology (RST), and the first two prototype exterior vehicles are now undergoing testing [5, 6].
Figure 3. The MDARS Interior robot with the new
security sensor package developed by Cybermotion, Inc., Salem, VA.
While previously implemented security robots such as ROBART I & II and MDARS have been capable of detecting and assessing a variety of security threats (i.e., intruders, fire, flood, chemical agents), they have not incorporated any capability for countering the actions of a confirmed intruder. In fact, some form of non-lethal response is a mandated Phase-II objective of the MDARS program, and a varied assortment of proposed solutions are being investigated.
Figure 4. The MDARS Exterior platform, under
development by RST, Inc., Westminster, MD.
The benefit of mounting a weapon on an unmanned vehicle to reduce risk to human personnel is obvious. As early as 1982, the Army's Missile Command (MICOM) began investigating robotic systems with (lethal) weapons for battlefield use. The initial focus was on allowing a soldier to remotely fire a vehicle-mounted anti-armor missile at a tank instead of having to carry the missile on his shoulder. The first prototype of the Grumman Robotic Ranger was fabricated in 1984, and remote missile firing was demonstrated. In 1985, remote missile and machine gun firings were also demonstrated from the RDS Prowler.
These successful demonstrations led to the formulation of the Teleoperated Mobile Anti-Armor Platform (TMAP) program, and prototype systems were procured in 1987/1988 from Grumman and Martin Marietta. Both systems were joystick-controlled over a fiber-optic tether, with the operator navigating via the returned TV image. Unfortunately, congressional direction in December 1987 prohibited the emplacement of lethal weapons systems on robots, and TMAP was redirected towards a RSTA mission and subsequently renamed the Teleoperated Mobile All-Purpose Platform [7].
The Army's two TMAPs were joined by the TeleOperated Vehicle (TOV), developed by NOSC under the Marine Corps Ground/Air TeleRobotic Systems (GATERS) program, in a live-fire demonstration at Camp Pendleton in September 1989, designating targets for laser-guided Hellfire missiles and Copperhead rounds. The TOV system was built on the HMMWV vehicle and featured an advanced telepresence control system, a high-bandwidth fiber optic link, and RSTA sensor package mounted on a telescoping mast (Figure 5). An earlier demonstration of the TOV had also featured the successful remote firing of a .50-calibre machine gun [8].
Figure 5.
The USMC TeleOperated Vehicle (TOV) developed by NRaD employed a remotely actuated
.50-calibre machine gun.
The end of the Cold War has triggered increased military interest in a variety of non-lethal weapons for application to such emerging focus areas as Law Enforcement and Operations Other Than War (i.e., peacekeeping, drug interdiction). Mounting a human-supervised non-lethal weapon on an unmanned vehicle can serve a variety of purposes, such as: (1) self-defense of the vehicle itself as a military asset; (2) prevention of destructive or hostile acts by individuals or groups of individuals; and/or (3) detention of individuals until human reinforcements can arrive to take them into custody. Any given weapon system will achieve these purposes by employing some mix of incapacitation and intimidation.
ROBART III
is intended to be an advanced demonstration platform for non-lethal response measures, effectively extending the reflexive teleoperated control concepts developed on ROBART II into the realm of coordinated weapon control. The basic issue to be addressed is the relative difficulty associated with effectively controlling a mobile robot equipped with a surveillance and/or targeting camera and an articulated weapon system.Experience gained through actual use by law enforcement personnel of conventional teleoperated devices of this type has revealed considerable shortcomings from a man/machine interface point of view. Simply put, if a remote operator has to master simultaneous manipulation of three different joysticks (i.e., one for drive and steering, another for camera pan and tilt, and yet a third for weapons control), the chances of hitting a moving target are minimal. In actuality, the single task of simply driving a teleoperated platform using vehicle-based video feedback is no trivial matter, and can be stressful and fatiguing even under very favorable conditions [8, 9].
Easing the driving burden on the operator was a major force behind the development of the reflexive teleoperated control scheme employed on ROBART II [10]. The robot's rich array of collision-avoidance sensors, originally intended to provide an envelope of protection during autonomous transit, were called into play during manual operation as well to greatly minimize the possibility of operator error. The commanded speed and direction of the platform was suitably altered as needed by the onboard processors to keep the robot from running into obstructions.
From these elementary beginnings, there eventually evolved a fairly sophisticated high-level supervisory control strategy wherein the operator could easily control the platform in manual mode by clicking on special behavioral icons depicted on the teleoperation display screen [2]. For example, selecting a "wall" icon to either side of the robot's own icon would cause the platform to enter wall-following mode, maintaining its current lateral offset from the indicated wall using side-looking sonar data. Choosing the "door ahead" icon in front of the robot caused the system to seek out and negotiate a perceived opening dead ahead, while the "door left" or "door right" icons could be used to turn and enter the next doorway or opening encountered on the indicated side of the path. In all cases, the on-board driving camera would be automatically pointed in the correct direction at the appropriate time to afford the remote operator the proper field of view.
The innovative thrust behind the ROBART III development is an extension of this reflexive teleoperated control concept one step further, incorporating sensor-assisted weapon control into the integrated package. The philosophy basically involves first making two of the three controllable elements (i.e., drive control, camera control, and weapon control) slaves to the third, so the human operator only has to deal with one entity. For example, the surveillance camera can be slaved to the weapon, so that the camera looks wherever the operator points the gun. If either the weapon pan-axis controller or the camera pan-axis controller approach their respective limits of allowable travel, the robot's drive controller causes the mobility base to rotate in place in the proper direction to restore the necessary range of motion. Alternatively, the weapon could be slaved to the surveillance camera, and so forth. In all cases, final closed-loop control of weapon pan-and-tilt can be provided by the video motion detector and other on-board sensors.
The various subsystems employed on ROBART III can be lumped into four general categories for purposes of understanding their interaction in support of an integrated reflexive weapons control system: (1) mobility and navigation; (2) threat detection and assessment; (3) non-lethal response; and (4) high level control. Each of these will be briefly discussed below.
ROBART III is intended only for indoor operation on relatively smooth planar floor surfaces. Differential steering is employed with a single passive caster in the rear of the platform directly behind the battery compartment. The left and right drive wheels are 8-inch wheelchair snow tires (see Figure 6) driven by a pair of 12-volt motors identical to those used on ROBART II. High-resolution phase-quadrature incremental optical encoders are attached to the motor armature shafts for precise velocity control and dead-reckoning displacement information. A three-axis Precision Navigation TCM Electronic Compass Module provides magnetic heading, temperature, and vehicle attitude (pitch and roll) information. System power is supplied by a 80-amphour 12-volt gel-cell battery which provides for roughly 6 to 10 hours of continuous operation between charges.
A combination of Polaroid ultrasonic sensors and Banner diffuse-mode optical proximity sensors are strategically located to provide full collision-avoidance coverage in support of the advanced teleoperation features desired. A 16-channel sonar multiplexer based on the bi-directional LH1500 solid-state relay is used to sequentially select individual transducers for connection to a single Polaroid 783821 ranging module. Only seven sonar transducers have been installed to date: two head-mounted sensors (see again Figure 6), and a five-element forward-looking array on the front panel of the mobility base (not shown). Additional transducers will be located on each shoulder pod, the front body trunk, and the sides and rear of the robot as the associated fiberglass body panels are fabricated. Two Banner SM312D near-infrared proximity sensors are installed on top of the head for collision avoidance purposes, while a longer-range SM912D unit is located behind the face plate, intended primarily for locating the left and right edges of open doorways. Additional SM912D proximity sensors will be installed on the front and sides of the vehicle in the near future.
Further enhancement of the autonomous navigation capabilities developed on ROBART II and later on the MDARS Interior robot are not a high priority for the near term on ROBART III, as the major focus will initially be on improved reflexive teleoperation. For this reason the existing world modeling and path-planning algorithms employed on these earlier units will be adapted to suit the preliminary needs of this effort. None of this integration has yet taken place at this stage of the development.
Figure 6.
Shown here with the shroud and front access panel for the mobility base removed, ROBART
III is a differentially steered indoor security robot approximately 42 inches high and 25
inches wide at the shoulders.
Similarly, further improvement of the intrusion detection and assessment subsystem over that currently employed on the MDARS Interior robots is not being pursued as part of this project. Extensive efforts were previously invested along these lines in conjunction with ROBART II, resulting in considerable technology which was subsequently transitioned to Cybermotion, Inc. in the 1992-1994 timeframe under a Cooperative Research and Development Agreement (CRADA) with NRaD [11]. Cybermotion in turn produced a highly optimized packaging of the sensor suite, dubbed the Security Patrol Instrumentation (SPI) module, for commercial sales in conjunction with their K2A Navmaster vehicle [12, 13]. This prototype SPI module was even further improved under a subsequent MDARS contract to the configuration shown earlier in Figure 3.
In order to support the development of an integrated non-lethal response capability, however, some rudimentary form of detection and assessment capability was required for ROBART III. Initial detection is accomplished by a 360-degree staring array of eight passive-infrared (PIR) motion detectors arranged as a collar ring just below the head. The head pans to the center of any alarmed zone, whereupon a miniature black-and-white video surveillance camera surveys the scene, and any subsequent motion is detected using a reconfigurable video line digitizer developed in conjunction with ROBART II [2]. Additional head-mounted security sensors (i.e., two Polaroid sonar transducers, the Banner SM912D proximity sensor, and an AM Sensors microwave motion detector) support the intelligent security assessment algorithm in rejecting nuisance alarms [12].
While other schemes such as explosively deployed sticky nets or chemical agents could have been employed, the non-lethal-response weapon chosen for incorporation on ROBART III consists of a pneumatically powered dart gun capable of firing a variety of 3/16-inch diameter projectiles, including tranquilizing darts. For purposes of feasibility demonstration, simulated tranquilizer darts were employed to illustrate a potential response application involving remote firing of temporarily incapacitating rounds by law enforcement personnel. One preliminary version of a test dart consists of a sharpened 20-gauge spring-steel wire approximately 3 inches long and terminated with a 3/16-inch plastic ball. A rotating-barrel arrangement (Figure 7) is incorporated to allow for multiple firings (six) with minimal mechanical complexity. (The spinning-barrel mechanism also imparts a rather intimidating psychological message during system initialization.)
Figure 7.
The six-barrel Gatling-gun arrangement facilitates multiple firings with minimal
mechanical complexity.
The darts are expelled at high velocity from their 12-inch barrels by a release of compressed air from a pressurized accumulator at the rear of the gun assembly. To minimize air loss, the solenoid-operated valve linking the gun accumulator to the active (top) barrel is opened under computer control for precisely that amount of time required to expel the projectile. The valve assembly is a modified dishwasher fill valve, bored out for minimal flow restriction and rewound for 12-volt DC operation. The gun accumulator is monitored by a MicroSwitch model 242PC150G electronic pressure transducer, and maintained at a constant pressure of 120 psi by a second solenoid valve connected to a 150-psi air bottle externally mounted on the rear body trunk. In addition to single-shot mode, all six darts can also be fired in rapid succession (i.e., approximately 1.5 seconds) under highly repeatable launch conditions to ensure accurate performance. The main air bottle is automatically recharged by a small 12-volt reciprocating compressor mounted in the robot's base, and monitored by a Data Instruments model EA200 pressure transducer.
The rotating barrel assembly is powered by a miniature PortEscap gearhead motor with an armature-driven phase-quadrature optical shaft encoder. A Banner SM312FP proximity sensor (see Figure 7) is fiber optically coupled to look down the bore of the bottom barrel to determine the presence or absence of a dart. Before the weapon is loaded, the gun encoder is initialized by slowly rotating the barrel assembly under computer control until a reflection/no-reflection transition is sensed, indicating the presence of an empty barrel. Once this initial referencing operation is complete, the computer can precisely align each barrel with the valve orifice by indexing a predetermined number of encoder counts in the clockwise direction. The system can also determine and track how many rounds have subsequently been loaded and/or fired using the same sensor.
Azimuthal and elevation information from the motion detector is available to the right-shoulder pan-and-tilt controller for purposes of automated weapon positioning. To facilitate aiming the weapon in manual mode, a 5-milliwatt 670-nanometer (visible-red) laser is bore-sighted to the dart gun barrel. (For covert operations, a near-infrared laser module can be substituted, as the video camera was specifically selected to be sensitive in the near-IR spectrum.) Watching video relayed from the head-mounted surveillance camera, the remote human operator simply aims the gun with a joystick until the laser spot is observed on the desired target.
The seven low-level microprocessor-based controllers currently installed on ROBART III will be interfaced via a high-speed multi-drop RS-232 interface as slaves to a master 68HC11 controller that serves as the functional equivalent of the Scheduler computer on ROBART II and the MDARS robots [2, 5]. The Scheduler in turn will be in full-duplex communication with the 486-based desktop PC host control station via a 9600-baud Telesystems spread-spectrum RF link. Initially all high-level planning functions will occur on the PC in similar fashion to ROBART II, but plans are to eventually install an onboard PC-104 stack to reduce the required RF throughput and significantly improve local processing capabilities.
Although construction of ROBART III began in 1992, progress has been relatively slow as the project is unfunded and all work is being accomplished at night and on weekends as time permits. Most of the effort to date has concentrated on the construction of the various mechanical assemblies and actuators associated with the weapon, mobility base, and upper body components, to include the head pan axis, right shoulder pan and tilt, and left shoulder pan. In addition, all low-level assembly-language software drivers have been written for the various PID motor controllers, and the video motion detector interface has been implemented. The pan action of the head can be automatically controlled by the video and passive-infrared motion detector outputs to track a moving target.
Considerable emphasis is being placed at this stage of the design process on safety issues, for obvious reasons. With any robotic system, there is always the danger of unexpected actuator engagement that must be taken into account. This is particularly true when designing a highly modular system where there is the possibility of multiple states arising as individual subsystems are powered up and down, and perhaps even temporarily disconnected. All outputs should default to the off position under all possible conditions, to eliminate surprises. When a weapons system is involved, however, such considerations become critical as opposed to merely prudent.
Accordingly, the weapon firing solenoid driver incorporates a pair of control lines, one active-high and the other active-low, to minimize the chances of premature actuation when the microprocessor's I/O port is powered up but not yet initialized, or alternatively when the computer is intentionally or inadvertently reset. The same dual-line strategy is employed on the dedicated "Weapon Fire" control link running between the Scheduler CPU and the weapons controller. In addition, both a "Weapon Ready" and a "Weapon Arm" command must first be sent down the RS-232 serial link to the weapons controller in the proper sequence before the "Weapon Fire" link will even be recognized. Additional efforts are ongoing to harden the system against possible vulnerabilities that may arise under abnormal operating conditions or configurations.
Fabrication of the remaining fiberglass body panels and the installation of their associated ultrasonic and near-infrared collision avoidance sensors is also in progress. The next step in the development involves incorporating the H-bridge power amplifiers in the mobility base, and integrating with the drive and steering controller and associated dead-reckoning algorithms. This will be followed by installation of the onboard master controller (Scheduler) and final development of the reflexive weapon control strategy, wherein the head pan axis, weapons pan-and-tilt control, and vehicle heading will be smoothly integrated into a seamless package that can be easily supervised by a remote operator.
ROBART III serves as a demonstration testbed for the incorporation of a non-lethal response capability into a body of robotic mobility, navigation, and threat detection and assessment techniques developed over more than a decade of research into the issues associated with site security robots.
The system will be able to confront intruders with a laser-sighted six-barrel tranquilizer dart gun, remotely controlled by a human operator. The system must simplify the operator's difficult coordination task of simultaneously controlling vehicle movement and attitude, pointing the camera, and aiming and firing the weapon. This will be accomplished by onboard intelligence which uses the inputs of various sensors including video motion detection to detect intruders, filter out false alarms, and automatically track a moving target.
Check out and initial integration of the various subsystems of ROBART III are currently in progress, but constrained by limited project resources. For further updates, see the NRaD web pages at http://www.nosc.mil/robots/index.html.
SPIE Mobile Robot and Automated Vehicle Control Systems, Boston MA, 20-21 November 1996, Volume 2903
Robotics at Space and Naval
Warfare Systems Center, San Diego
Last update: 9 February 1998.
robo-web@spawar.navy.mil