Lieutenant Colonel John Blitch (ret.), director of the Center for Robotic Assisted Search and Rescue (CRASAR), requested a team of robot experts and suppliers to assist in search efforts at the World Trade Center (WTC) disaster site. New York City’s Office of Emergency Management requested CRASAR’s response directly. Blitch also had an understanding with Special Operations Command Deputy Chief Ray Downey to respond if needed. CRASAR is a nonprofit organization based in Littleton, Colorado, and was called to assist in the rescue efforts at the WTC using any available robots within hours of the September 11 disaster. Responders included Defense Advanced Research Project Agency (DARPA); U.S. Army Tank Automotive and Armaments Command-Army Research and Development Command-Explosive Ordnance Disposal Division (USA TACOM-ARDEC-EOD); Dr. Robin Murphy, director of the Perceptual Robotics Laboratory, University of Southern Florida (USF), Space and Naval Warfare Command (SPAWAR); and Foster-Miller and iRobot, robot manufacturers.


All team personnel assembled at Stewart Air Force Base in Newburgh, New York, where Phase I personnel were immediately called forward to the Federal Emergency Management Agency (FEMA) base at the Jacob Javits Convention Center in New York City. They stowed their personal gear and went to Ground Zero early Wednesday morning. They spent the morning outlining the organizational structure and familiarizing themselves with the immediate surroundings.

The CRASAR team arrived at Ground Zero on September 12 and shortly thereafter began the first phase of the operation, search and rescue. Working directly with the Fire Department of New York (FDNY) and FEMA teams, Foster-Miller’s and Inuktun’s robotics systems supplied video and audio support in confined space searches. Foster-Miller personnel operated the robots for the Phase 1 operations. Together with Dr. Robin Murphy and graduate students from USF, they also provided technical support and advice in robotics.

Foster-Miller’s Talon about to enter a void. (Photos courtesy of Foster-Miller.)
Click here to enlarge image

Operations occurred on a 24-hour basis, often without a break. Personnel operated for between 12 and 24 hours a day during the first phase.

Initially, Lt. Col. Blitch met with FDNY officials, and Foster-Miller oversaw the field operations with two teams. One team was assigned to support FEMA Indiana Task Force 1 and accompanied personnel on their shifts. The other team functioned as a freelance operator, providing support to rescuers as requested, primarily FDNY and FEMA Pennsylvania Task Force 1. Arnis Mangolds of Foster-Miller was the team leader, acting as the field interface with the battalion chiefs and as backup operations manager.

During the first phase, the team used the smaller Foster-Miller Special Operations Lemming (SOLEM) and Inuktun VGTV (Variable Geometry Tracked Vehicle) and MicroTrac robots. Other robots were available (Foster-Miller Talon and iRobot), but for this phase they were too large for the confined workspace. Other robots offered by DARPA included the U.S. Navy SPAWAR systems and iRobot PACBOT, but because of their large size, fragile nature, and limited mobility, they only performed practice runs at a nearby abandoned parking garage.

The Talon’s camera view as it descends the stairs into the WTC’s subbasements.
Click here to enlarge image

The U.S. Army Night Vision Laboratory provided thermal imaging cameras, which, unfortunately, could not withstand the high temperature of the fires in the debris. Since the robots had adequate onboard lighting, there was no need for infrared (IR) cameras and, although on standby, these cameras were not used.

Phase I robotics personnel had no idea of who was in charge, whom to report to, or how to fit in. As a result, initially, Phase I personnel used top-down and bottom-up approaches for operations. In the top-down approach, they received the proper credentials but no assignments. In the bottom-up approach, the team would either match up with a FEMA team (such as Indiana Task Force 1) and accompany it on shifts as a dedicated asset or act as a free roamer, reporting to the on-pile section and battalion chiefs. The teams were not used to working with us and were not fully aware of our capabilities.

The robots were deployed using operator control units (OCUs) that included viewing displays that allowed multiple audience viewing and recording. The SOLEM used a wearable backpack unit and the Inuktun MicroTrac and VGTV robots used a lunchbox-sized system. The Inuktun OCU, however, had a separate display. Both OCUs were designed for all-weather use, had instant-on, and allowed for highly mobile transport and setup. All controls were proportional for easy and intuitive operation. Operators required very little training, and when TACOM-ARDEC-EOD teams arrived in Phase II, the training-to-operation time required was less than an hour. The Talon unit and the first SOLEM unit were radio-controlled; all the others used durable 100-foot tethers to interface with the OCUs. Large amounts of metal debris interfered with the SOL-EM’s and Talon’s radio control signals. The tethers tended to snag on debris.

Camera view of slurry wall during inspection.
Click here to enlarge image

The robots’ missions included searching voids for human remains, confined space reconnaissance, and monitoring structural integrity. When remains were found, the spot was marked for later recovery by search personnel. The Talon’s gripper arm enabled it to retrieve victims’ personal items and move debris aside.

In confined space reconnaissance, the units explored passageways for structural integrity and safety for entry search and rescue personnel. Later, in Phase II, attached sensors monitored void atmospheres.

Most searches in Phase I were shallow. The robots were used initially to extend the limits of flashlights, dogs, and pole cameras in searching for victims; in Phase II, they became equally important for determining the extent and condition of the passageways. Identifying dead-ends quickly allowed reallocation of time-critical resources to finding openings, larger openings, or victims. Although the robots initially were used in the hope of the “big score” of searching whole floors in surrounding buildings, human teams performed these tasks. They liked doing it, were trained for it, and were better and faster at it.

Robot movement was difficult because of obstructions, potential shifting debris, and the ever-present danger of voids. Everything had to be backpacked into and set up on often-precarious positions. The robot’s mobility had to be completely rethought. There were virtually no flat, horizontal surfaces. High centering on twisted metal knife edges was the rule. Rebar caught on the tethers and skewered tracks. Dust and loose paper on the surfaces hid what was underneath and made path planning difficult. Pathways were three-dimensionally tortuous. From an operational point of view, it was highly daunting and, in fact, at first appeared hopeless. There were no flat, horizontal stretches—nothing but jagged edges and, apparently, insurmountable obstacles. The robots had not been designed for that kind of operation. The dust and smoke were distracting but not a problem. The heat, however, was: As the vehicles penetrated deeper into the pile, the intense heat melted the tracks. When the Inuktun was retrieved from the pile for the final time, it was on fire.

The greatest problem for the operators was recognizing what the cameras were seeing. Looking through the narrow field of view of a camera can be difficult under the best of conditions, but the three-dimensional nature, with no flat surfaces; the crushed, torn, twisted, and distorted debris; and the thick dust covering made interpreting the images difficult. Trained firefighters had to help. The vehicles were often lowered in by the tether and were free hanging. Not knowing which way was up further complicated interpretation.

A team of structural engineers directed the inspection process and interpreted the data in real time. Considering the severity of distortion, as many eyes on the scene as possible were needed. Often, more than a dozen inspectors, health officials, and firefighters would watch and direct the progress of one robot. The camera views were recorded and passed to the inspectors for later review and permanent recordkeeping. Often the same area had to be reinspected as secondary collapses occurred and underground fires progressed.

On September 17, during operations with the Indiana Task Force, the radio-controlled SOLEM unit was caught by rebar in a cavity and could not be retrieved. A rope had been attached to it to facilitate retrieval, since metal debris sometimes interfered with the unit’s radio signal. A nearby fire prevented any attempts to recover the system using personnel. The rope broke, and the unit plunged to a lower level.

Foster-Miller personnel returned to the company’s Massachusetts headquarters and constructed a new unit from spare parts. The rebuilt SOLEM-2 had extra dimmable lights, camera pan-and-tilt (the original had just a tilt camera), tether control, and wider tracks. The modifications made it more suitable for operations, and it was heavily used over the next two weeks.

Operations ran 24 hours with work schedules arranged in 16-hour overlapping shifts. Initially, the team leaders would report to the battalion chiefs to receive assignments but later, once the capabilities were identified, the teams were called on the fly. Bucket lines would be formed into the debris pile, supplies were stockpiled, and other assets such as dogs or technical equipment were spread throughout. As a need arose, the firefighters would call back for additional tools, supplies, or technical assets such as the robot systems. After the immediacy of the rescue operation changed to one of recovery, the robots were used for deeper and longer operations, and the operations were generally planned in advance.

When the robotic devices’ capabilities and mission profiles were determined, other user groups wanted to take advantage of their capabilities.

Foster-Miller, USF, and TACOM-ARDEC-EOD were the only robotics personnel employed at the WTC.


The majority of Phase I personnel pulled out September 20 and were replaced with TACOM-ARDEC-EOD personnel, who were trained to operate the robots. Foster-Miller personnel remained as technical advisors until October 2. As the weeks progressed, the emphasis shifted to recovery and structural analysis.

Lt. Col. Blitch felt it best to get the Army involved for two reasons: TACOM-ARDEC-EOD’s familiarity with most of the robotic platforms being used at the WTC and the danger of using untrained personnel in this environment. He requested personnel familiar in robotic support from TACOM-ARDEC-EOD through the Department of Military Support on September 16. All soldiers deployed to the WTC were assigned to TACOM-ARDEC-EOD, Picatinny Arsenal, New Jersey, and were explosive ordnance disposal technicians and combat lifesavers and trained in confined space operations.

Once TACOM-ARDEC-EOD arrived on September 23, the operation’s focus had transitioned into recovery and structural integrity stages. I led the Army team; we reported to Colonel Conners, who commanded all U.S. Army personnel on the site and received assignments from City of New York Department of Design and Construction (DDC), FDNY, and the Port Authority of New York and New Jersey Police Emergency Services Unit (PA ESU).

On September 25, the Army team became the lead robot operators, interfacing with the fire officials; Foster-Miller, which represented CRASAR, provided technical support. Most missions occurred along the Liberty Street area.

Phase II robotics personnel received requirements from three different agencies: the DDC, FDNY, and PA ESU. Once requirements were identified and prioritized, the limited assets were assigned. This approach required an enormous amount of perseverance, as acceptance was neither immediate nor universal. Once proven, however, we were constantly on the move responding to on-the-spot assignments, fast-action quick response assignments. As time to accomplish the mission was extremely short, this approach made the best use of limited time and assets. The pace of the search operation combined with the ever-present physical dangers meant that, if the equipment was not ready the instant it was needed, it was not used and the operation moved on. One-person teams were very mobile, could respond quickly, and fit in well with the overall operation. Two-person teams were the maximum useful size.

Determining structural integrity, particularly of the retaining wall around the WTC site, was a primary mission. The robots’ cameras enabled viewers to view the wall, identifying damage or threats to its stability, such as debris leaning up against it. The DDC and U.S. Army Corps of Engineers helped direct some of the operations. A major advantage of the robots was that many eyes could view, interpret, and direct subsequent action in real time. The robots were used for illuminating and viewing the integrity of the slurry wall and its supports. The structural engineers opened up an access hole and the vehicles were lowered in. Sometimes the vehicles were lowered and swung onto a platform from an overhanging position. The vehicles had to withstand impacts with the concrete walls.

As underground fires shifted and spread, firefighters and NIOSH investigators attached environmental probes to the vehicles to monitor atmospheric conditions as the robots performed their missions. The atmospheric probes were used before personnel entered confined areas. Strapped on the vehicles were probes that assessed carbon monoxide, hydrogen sulfide, volatile organics, temperature, and structural integrity.

By October 1, the mission backlog was completed. TACOM-ARDEC-EOD and Foster-Miller personnel were released on October 2.


Waterproofing. This became one of the most important features. Because of blood-borne pathogens, everything leaving the Ground Zero work site had to be decontaminated and rinsed. This became a serious factor on October 1 after we pulled the last operable Inuktun out of a hole; its exposed wiring had caught fire from the water being pumped onto the debris. Some robots had to be carefully handled because they were not waterproof, which added to an already long day.

Durability. The robots were treated roughly. During some missions, they were lowered down 60 to 70 feet and swung onto landings of the WTC complex. While driving robots on the pile and down the center of I-beams, it was not uncommon for the track to melt. On two of the robots, the nylon gears that controlled the cameras tended to strip. Talon and SOLEM-2 were the only two robots to last the entire mission without a major repair.

Awareness of robotic capabilities. On many occasions, search and rescue personnel were not aware of the availability of robotics at the site. For example, search and rescue personnel would see a wall of rubble and could not proceed. When we arrived on-scene we saw pipes, air-conditioning ducts, and other openings where our robots could easily be employed.

Durable tether. The tether had to be waterproof and strong enough to handle the weight of the robot being lowered or dragged.

Cost. The robot costs range from $16,000 to $60,000, depending on capability; but this cost is negligible when it keeps a human from harm’s way unnecessarily.


It is my hope that in the future robots are used not only in bomb disposal and search and rescue operations but in all areas where humans are put in harm’s way such as SWAT operations, hostage negotiations, and hazardous spills.

DAVID PLATT is a sergeant first class with the United States Army and is assigned to the Tank Automotive and Armaments Command-Army Research and Development Command-Explosive Ordnance Disposal Division (TACOM-ARDEC-EOD) at Picatinny Arsenal in New Jersey.


Four robot units were used for victim search, void exploration, and structural integrity reconnaissance at the World Trade Center site.

Foster-Miller Talon. This 75-pound radio-controlled robot, designed for reconnaissance, breaching, and recovery operations, has all-weather, day-night, amphibious, and stair-climbing capabilities. It includes a quick-release cargo bay that accommodates a variety of sensor payloads and can carry a load of up to 300 pounds. The Talon carries four cameras and sets of lights and can be outfitted with forward-looking infrared (FLIR), night vision goggles (NVG) and other sensors and grippers and saws on its 53-inch double-jointed arm. With stair-climbing capabilities, Talon is intended for deep exploration in buildings, basements, and subway systems.

Foster-Miller SOLEM (Special Operations LEMming). This 33-pound tethered lightweight robotic rover is designed for field reconnaissance. It is a highly mobile platform with all-weather, day-night, and amphibious capabilities. The SOLEM mast was fitted with a special head to illuminate the target area with a laser grid from which to take physical measurements. The Laser Unexploded Ordnance Reconnaissance (LUXOR) head was equipped with daylight and nighttime cameras. The head could be swapped out in the field for other camera types, but during operations the LUXOR head was used. The SOLEM was fully equipped with radios and batteries, and its wide track provided good traction. However, the drive motors were not always powerful enough to handle climbing 80-degree terrain. The eight-inch body height and 10-inch pan-and-tilt arm provided good visibility over debris. Previously, the SOLEM was used in Bosnia as a spotter and assistant in moving ordnance.

Inuktun VGTV (Variable Geometry Tracked Vehicle). This tethered device is used for small holes and voids, fits into a backpack, and weighs less than 10 pounds. A 30-pound battery pack carried in a backpack provides external power. It has one camera and is not waterproof. The control box is interchangable with other Inuktun robots. It is small and has a user-friendly operator control unit (OCU) and shape-shifting capability. The treads can be reconfigured to adjust to the terrain requirements, and all system components can be put in one backpack. However, it is not rugged or waterproof, has exposed parts, is not very mobile, uses nylon gears for the camera (which tend to strip from rough handling and conditions), throws tracks easily, and has a screen display separate from the OCU.

Inuktun Microtrac. This tethered, nonwaterproof, battery-operated device is used for tiny to small holes and voids. It can be backpacked. It has one camera, is small and lightweight, and has a user-friendly OCU. All system components can fit into a backpack. It, however, is not rugged, has low clearance, and uses easily stripped nylon gears to control the camera; also, the display is separate from the OCU.

No posts to display