NIST’s Emergency Response Robots Project, led by Adam Jacoff, validated 20 standard test methods for evaluating aerial drone capabilities and remote pilot proficiency at the Unmanned Tactical Applications Conference. The event was hosted by FlyMotion at the Guardian Centers responder training facility in Perry, GA on October 4-8, 2021. The NIST team, which includes emergency responders and test facility managers, helped familiarize more than 120 participants from all over the world with our standard test methods and related training scenarios embedded with quantitative measures of performance.
These standard test methods have been developed by NIST with support from the Science and Technology Directorate of the U.S. Department of Homeland Security (DHS). They are being standardized through the ASTM International Standards Committee on Homeland Security Applications; Response Robots (E54.09).
Our process for developing, validating, and standardizing a growing suite of more than fifty test methods for ground, aerial, and aquatic systems, includes several types of events. We host requirements workshops with emergency responders, validation testing with robot manufacturers, competitions with robotics researchers, and standards committee meetings with all the above. We also conduct inter-laboratory experiments with other facilities that host our test methods for their own evaluations. For all these various communities, our standard test methods provide a tangible language of robot tasks that enable quantitative measures of robot capabilities and remote operator/pilot proficiency. They help guide innovation, measure progress, and highlight breakthroughs that everybody can understand and appreciate.
This event also included tactical scenarios demonstrated live with actors as perpetrators, victims, and emergency responders. They showed how drones and other robots can be incorporated into fast-paced and chaotic emergency response operations including simulated gun shots, fires, and smoke.
After each scenario demonstration the NIST team embedded standard test apparatuses to guide drone pilots through a sequence of essential tasks resulting in a quantitative score. These embedded test apparatuses enable performance of drones and pilots to be measured, compared, and tracked over time, which was never possible in such complex scenarios.
A variety of firefighting, public safety, and search & rescue scenarios were conducted:
The Open Test Lanes evaluates 5 different flight paths to identify objects from safe altitudes in open environments. These tests are scalable for all sizes of aircraft to demonstrate positive control at all times with accurate perches. They can be performed outdoors or indoors to control lighting and weather. The smallest size lane fits on an indoor basketball or tennis court for small drones and/or novice pilots to practice safely without flying in the national airspace.
The Open Test Lanes were located near related scenarios for wide area search of a plane crash site and a vehicle identification scenario. Two were set up with 3 m (10 ft) spacing between omni bucket stands and one was set up with 6 m (20 ft) spacing to be operationally relevant for various scenarios. All were set up with alternating white and black bucket stands to evaluate remote pilot proficiency using the Payload Functionality variant of the tests rather than the basic Maneuvering variant of the tests that use all white buckets throughout. The Payload Functionality tests involve a modest workload for the remote pilot to pantilt-zoom, focus, and control exposure at each bucket for scoring after the trial. The alternating white and black buckets force maximum exposure control.
Open Scenario: Wide Area Search
Quad Screen Trial Video Example - Wide Area Search
The Open Test Lane evaluates flight paths to identify objects from safe altitudes in open environments. Related scenarios include wide area searches such as this simulated plane crash site that used all the same omni bucket stands from the Open Test Lane. There are 20 targets overall, each with 5 increasingly small features to identify for a total of up to 100 points available for a complete trial. This enables comparison of scores for pilots and aircraft that can reliably perform the various bucket alignments and identify the smallest visual/thermal acuity features across all available acuity targets. The trial time limit is typically set to 20 minutes to remain within one battery charge and to maintain a schedule throughout the day for multiple pilots. Time limited trials also enable direct comparisons of scores for completeness and efficiency. Only scores using similar aircraft and trial times are directly comparable to evaluate pilot proficiency, but a variety of different aircraft can be used to compare overall scores and ease of use.
Open Scenario: Vehicle Identification
Quad Screen Trial Video Example - Vehicle Identification
(Scenario layout training: find the errors in the sticker placements around the vehicle)
The Open Test Lane also leads pilots toward vehicle identification scenarios from safe hover altitudes above nearby obstacles. The designated flight path around the vehicle includes an orbit with equal radius and altitude to align with the 45-degree angled omni buckets on the roof (bucket targets A1, B1, C1, D1). The chosen orbit radius and altitude is based on the height of surrounding obstacles, the intended mission requirements, the aircraft’s zoom capabilities, etc. Any orbit can be used, but only trials with similar orbits and trial times are comparable. There are also precise perch positions and orientations on the road directly under the front and rear orbit positions. These perches evaluate landing accuracy along with the functional pan-tilt-zoom capabilities of the aircraft while landed at the chosen orbit radius. Perching demonstrates the aircraft’s capability to maintain surveillance while conserving battery. The perch targets are buckets under the vehicle that represent operationally significant underbody objects in shadow (bucket targets A5, C5). Each side of the vehicle has 5 exterior visual/thermal targets to identify from the chosen radius, altitude, and perch positions. There are 20 targets overall, each with 5 increasingly small features to identify for a total of up to 100 points available for a complete trial. All targets are identified from the designated flight path starting with the angled buckets on the roof to verify the altitude and radius, then exterior features or surrounding ground objects, and exterior window targets to identify while presumably looking for interior objects. All targets are on the exterior of the vehicle to eliminate potential variations in scoring due to interior target obstructions, window glare, tinting, etc. Interior target identifications can be considered bonus points because they are less comparable across trials due to changes in sunlight.
The Obstructed Test Lanes enable remote pilots to fly safe and repeatable flight paths to inspect objects within close proximity to obstructions. They include a comprehensive set of 5 different tests that guide remote pilots through various standoff positions, orientations, and perches at 2-3 m (6-10 ft) from objects. They can be performed outdoors or indoors to control lighting, weather, and access to the Global Positioning System (GPS).
The Obstructed test apparatuses use “dual bucket alignments” to enable remote pilots to triangulate a safe standoff position by simultaneously aligning with a perpendicular bucket and its associated angled bucket. The dual bucket alignments form a right triangle with equal dimensions for the bucket separation and the aircraft standoff. They use pairs of 7.5 liter (2 gallon) buckets with 20 cm (8 in) diameter recessed targets inside at a 2 m (6 ft) spacing.
Alignment rings inside the perpendicular (90 degree) white buckets visually guide safe flight paths toward and away from the objects. Alignment rings inside the associated angled (45 degree) buckets indicate when the aircraft is in the designated safe and repeatable standoff position. That’s where a remote pilot can maintain position and operate their interface to identify increasingly small features on acuity targets inside the buckets. This modest operational workload includes zoom, focus, and exposure to capture images for scoring after the trial. White and black buckets require maximum control of exposure levels or thermal palettes to discern more details and score more points.
There are 5 different tests with increasing difficulty called Perch, Wall, Ground, Alley, and Post. The dual bucket alignments guide remote pilots through a series of 10 positions, orientations, and perches within both the standard test lanes and the operational scenarios embedded with scoring tasks. All tests and scenarios result in quantitative scores up to 100 points maximum to facilitate measurement, tracking, and comparison across different aircraft and/or remote pilots.
Procedure: The procedure is the same for all Obstructed tests. Each test has dual bucket alignments designated 1 – 4 that at are performed in a sequence. The sequence includes some backtracking to ensure the tasks are performed in various directions relative to the obstacles involved. The sequence of positions is 1 2 3 4 – 3 2 1 – 2 3 4 with the red underlined numbers indicating the backtracking part of the sequence. That results in 10 dual bucket alignments or 20 bucket alignments total.
Maneuvering Trials: A complete trial totals up to 100 points maximum for 20 bucket alignments. Points are scored using a single no zoom image of each bucket showing either a full alignment ring (5 points), a partial alignment ring (1 point), or no alignment ring (0 point).
Acuity Targets: All trials result in maneuvering scores. But Payload Functionality trials add an operational workload to identify acuity targets while aligned with buckets. The level of detail the system can discern should be known and set prior to conducting a Payload Functionality trial. Each acuity target has 5 increasingly small gap orientations to identify correctly. The smallest features are 1 mm (0.04 in) needed to read small text on shipping labels, for example. Each identifiably gap orientation is verbally conveyed to a Proctor during the trial, or when operating alone a single full zoom image of the acuity target can be captured to score after the trial.
Time Limits: Test trials are not intended to be races. But trial time limits can be imposed to minimize fatigue across multiple tests or to maintain a schedule. Trial time limits should be long enough for an “expert” to complete a perfect trial. Scores of incomplete trials due to expired time limits can only be compared to trials with the same elapsed time limits. Typical time limits are typically set to 5 minutes for Maneuvering trials and 10 minutes for Payload Functionality trials. Although any time limits may be used depending on the drone, the pilot, and the environment.
Faults: A fault is any contact with a test apparatus or any safety issue such as crossing a designated boundary between test lanes or the remote pilot flight line. Any fault results in an end of trial.
Metrics: A complete trial requires performing all the designated bucket alignments in order with no faults (contact with an apparatus) or safety issues (exceeding a boundary). If the trial is not complete, the metrics below do not apply. Keep practicing until complete trials are routinely achieved before applying these metrics:
Note: The verge points of the dual bucket alignments designate the most efficient locations to score all buckets. However, stable hovers at the verge points between buckets can be difficult to enforce similarly for various drone sizes and pilot proficiencies. So each bucket can be scored individually from any desired proximity, understanding that the resulting scores and trial times may be negatively affected.
Obstructed Scenario: Bus Hostage Situation
Quad Screen Trial Video Example - Bus Hostage Situation
The bus hostage scenario included role players as perpetrators and victims with actual law enforcement organizations from the region demonstrating simulated tactics including drones and a ground robot.
We added quantitative measures of performance to this scenario using dual bucket alignments to guide remote pilots safely through a designated series of inspection tasks. This enables comparison of scores up to 100 points for pilots and aircraft that can reliably perform the various bucket alignments and identify the smallest visual/thermal acuity features across all available targets. The trial time limit was set to 20 minutes to remain within one battery charge and to maintain a schedule for multiple pilots. Time limited trials also enable direct comparisons of scores for completeness and efficiency. Only scores using similar aircraft and time times are directly comparable to evaluate pilot proficiency, but a variety of different aircraft can be used to compare overall scores and ease of use.
The flight paths included safe and repeatable standoff positions, orientations, and perches within 2 m (6 ft) of objects for aircraft flying the most efficient flight path. But bucket alignments can be achieved from any standoff for larger aircraft using more capable zoom lenses. The embedded scoring tasks are 7 liter (2 gallon) buckets with 20 cm (8 in) diameter recessed targets inside. The perpendicular (90 degree) white buckets provide visual alignments for the remote pilot to trust as safe vectors to approach and leave the object being inspected. The associated angled (45 degree) buckets enable triangulation to maintain a safe proximity from the object while operating their zoom, focus, and exposure to capture images for scoring after the trial.
Each set of tasks included 10 positions and orientations worth 10 points each distributed throughout the scenario. The pilot scores 5 points for getting aligned with each perpendicular bucket plus another 1-5 points for correctly identifying increasingly small features on the acuity target located inside the associated angled bucket. The smallest features to identify are 1 mm (0.04 in) representing small text on shipping labels, for example. The white and black pairs of buckets require maximum exposure control to capture images for scoring after the trial.
Large vehicles like buses can require interior searches as well. These relate to the Confined Test Lanes described later, but are typically part of these types of scenarios. They are intended only for scoring by aircraft that can safely fly inside the bus.
Obstructed Scenario: Vehicle Takedown
Quad Screen Trial Video Example - Vehicle Takedown
Obstructed Scenario: Hazardous Tanker Truck Accident and Fire
Quad Screen Trial Video Example - Hazardous Tanker Fire
The tanker truck accident and fire scenario used 5 large bucket apparatuses to guide remote pilots to points of view around the vehicles for a fast tempo initial assessment. We also embedded 5 small bucket apparatuses near features needing detailed inspections such as the inside of the cab, dripping valves, and gauges to read. This scenario included drop accuracy tasks to emplace remote sensors trying to determine the nature of the hazard (see the sensors on the ground).
Obstructed Scenario: Structure Exterior Inspection
Quad Screen Trial Video Example - Structure Exterior
Structure exterior inspection tasks include looking through windows and doors along with surrounding ground objects of interest. In this case, the tasks were embedded around a partially collapsed structure. The objective was to safely fly in close proximity of about 2 m (6 ft) from the windows and doors to perform a window/door clearing maneuver with high/low and left/right views inside the structure.
Obstructed Scenario: Flooded House and Collapsed Bridge
Quad Screen Trial Video Example - Flooded House Scenario
Several houses were submerged with victims on the roof and others in the water nearby. The tasks to perform included delivery of personal floatation devices to those stranded on the roof, searching in windows for survivors, and identifying/tracking moving swimmers. The omni bucket apparatuses replaced the actors as objects to identify and the landing served as target for the dropped payload accuracy.
Obstructed Scenario: Night Operations House Surveillance and Search
This scenario was actually conducted the week prior at a different event, but nicely augments all of the scenarios above. This house surveillance at night used two sets of horizontal and vertical Obstructed test apparatuses on all four sides of the house, guiding the remote drone pilot to safe locations among a variety of very difficult obstacles such as the overhanging roof, trees, shrubs, power lines, shrubs, etc. A set of 5 vertical test apparatuses with 10 positions and orientations were attached to all sides of the house totaling 100 points. A set of and 5 horizontal test apparatuses with 10 positions and orientations were also placed on or near objects of interest around the house for another 100 points.
The Confined Test Lanes enable remote pilots to fly safe and repeatable flight paths to inspect objects within confined environments and interior room-to-room searches. They include a comprehensive set of 5 different tests that guide remote pilots through various standoff positions, orientations, and perches at 1 m (3 ft) from objects. They are half the size of the Obstructed Test Lanes but use all the same procedures and scoring. They can be performed outdoors or indoors to control lighting, weather, and access to the Global Positioning System (GPS).
The Confined Test Lanes and related scenarios use “dual bucket alignments” to enable remote pilots to triangulate a safe standoff position by simultaneously aligning with a perpendicular bucket and its associated angled bucket. The dual bucket alignments form a right triangle with equal dimensions for the bucket separation and the aircraft standoff. They use pairs of 1 liter (1 quart) buckets with 10 cm (4 in) diameter recessed targets inside at a 1 m (3 ft) spacing.
Alignment rings inside the perpendicular (90 degree) white buckets visually guide remote pilots along safe flight paths toward and away from the objects being inspected. Alignment rings inside the associated angled (45 degree) buckets indicate when the aircraft is in the designated safe and repeatable standoff position. That’s where a remote pilot can maintain position and operate their interface to identify increasingly small features on acuity targets inside the buckets. This modest operational workload includes zoom, focus, and exposure to capture images for scoring after the trial. White and black bucket pairs require maximum control of exposure levels or thermal palettes to discern more details and score more points.
There are 5 different tests with increasing difficulty called Perch, Wall, Ground, Alley, and Post. The dual bucket alignments guide remote pilots through a series of 10 positions, orientations, and perches within both the standard test lanes and the operational scenarios embedded with scoring tasks. All tests and scenarios result in quantitative scores up to 100 points maximum to facilitate measurement, tracking, and comparison across different aircraft and/or remote pilots.
Scoring: Each dual bucket alignment is worth up to 10 points, including 5 points for maneuvering and 5 points for visual/thermal acuity. Maneuvering points are scored using a single no zoom image of each perpendicular bucket showing either a full alignment ring (5 points), a partial alignment ring (1 point), or no alignment ring (0 point). Visual/thermal acuity points are scored using targets inside the angled buckets with 5 increasingly small gap orientations to identify correctly (1 point each up to 5 points). The smallest features to identify are 1 mm (0.04 in) representing small text on shipping labels, for example. Target identifications can be verbally conveyed to a Proctor during the trial, or a single full zoom image of each acuity target can be used to score the trial after landing. A complete trial includes 20 bucket alignments totaling up to 100 points maximum per trial.
Note: The dual bucket alignments define the most efficient positions to score all buckets placed around the object being inspected. However, stable hovers at the verge points between buckets can be difficult to enforce similarly for various aircraft sizes and pilot proficiencies. So each bucket can be scored individually from any desired proximity, understanding that the resulting scores and trial times may be negatively affected.
Confined Scenario: Collapsed Structure Search and Inspection
One of the confined scenarios at this event were semi-collapsed structures representing environments not safe for responders to search given the potential for further collapse due to aftershock or structural failures. The inspection tasks were placed around crushed vehicles within a collapsed structure. They were also distributed across individual rooms in a search pattern to ensure each room was inspected for victims and other objects of interest.
Confined Scenario: Subway Terrorist Attack
The subway terrorist attack scenario was in a simulated tunnel with multiple DC Metro railcars. A line of intact and upright railcars near the station entrance led to one railcar on its side almost blocking the tunnel. The scenario included several survivors inside and around the perimeter of the overturned railcar along with some suspicious packages that could have been secondary explosives.
Legged Mobility and Search Tests
Newly developed standard terrain tests for robots with legged mobility are based on plastic crates. They are quick and easy to reconfigure into increasingly difficult terrains. For example, simply as a flat terrain they have a grate-like surface that some autonomous robots (and animals) don’t see so well. Various terrains can be reconfigured including the Diagonal Hill (shown) or a Pyramid Hill with the highest point in the center of the room. Or every second crate can be flipped over in a checkered pattern to include negative obstacles or holes to avoid, forcing very specific footfall locations. These checkered terrains can be combined with the hill topologies for maximum difficulty. These crate-based terrain tests are also easy to purchase and fabricate all over the world. Note the white and black dual bucket alignment tasks all around the sides and top at different elevations and orientations. This is a quantitative way to measure and compare the search/inspect capabilities of robots while negotiating the terrains. The same search/inspect tasks can be placed over all the other standard terrains such as crossing ramps, sand and gravel crossover slopes, and stepfields, etc.
The Sensor Test Lane evaluates visual acuity, thermal acuity, hazmat label identification, motion detection, and latency while hovering at altitudes up to 90 m (300 ft). Also Survey Acuity for forward flying aircraft. It uses larger 120 liter (32 gallon) buckets with 45 cm (18 in) diameter inscribed rings to guide alignments. Panels with targets on white and black backgrounds point toward the hover position. Typical visual/thermal acuity targets are across the top line and operationally relevant objects are across the bottom line, including hazmat labels, partial license plates, gauges to read, weapons to identify, and thermal sources as examples. The back side of the target panels have QR codes of different sizes to capture in flyover video to measure the Survey Acuity of the system from various straight and level flight altitudes. The video can be postprocessed to automatically read as many incrementally small QR codes as possible for scoring. This can identify the smallest readable size across a large search area when multiple posters are laid out.