Army Research Laboratory Perception/Performance Evaluation Project

The detection, classification, and tracking of moving vehicles and people from an unmanned vehicle maneuvering at speeds of up to 80 kph remain a recognized risk to the successful development of Future Combat Systems (FCS) Unmanned Ground Vehicles (UGVs).  Current sensor and algorithm designs do not achieve FCS UGV objective requirements for safe operations in both urban and non-urban environments.  Ongoing development efforts are focused on the advancement of sensor and algorithm capabilities to enable safe operation of UGV’s in dynamic urban environments.  NIST conducts the research in support of these needed capabilities.  This work includes development and execution of algorithm and system technical evaluations and research leading to definition and use of appropriate evaluation metrics, measurement methods and calibration methods.

NST Capabilities that support research for ARL and others:

1. NIST provides independent expertise applied to a range of measurement tasks addressing the needs of Army robotics efforts, most recently with respect to perception elements.   Two of NIST’s core competencies are “Measurement Science” and “Rigorous Traceabililty.” These competencies have been applied to measurements important to the Autonomous Navigation System (ANS) and Collaborative Technology Alliance (CTA) human detection systems analysis exercises.
2. NIST’s Intelligent System’s Division has easy access through other NIST divisions and Laboratories to world class measurement support, guidance, and systems to support a full range of measurements. Such assistance has been valuable for such efforts as the calibration of the ANS sensor platform using the extremely precise API 6D LaserTracker and for precise optical surveys of test locations at GDRS Westminster and NIST.

API tracker used to precisely locate target points for evaluation of ANS sensor platform calibration procedure in NIST Mobility Laboratory. Tracker is in center of right image. 





3. NIST has developed, largely under internal and other project support, a position measurement system based on the Multi-Spectral Solutions, Inc. ultra wide-band (UWB) asset tracking system, suitable for tracking multiple objects over large areas.  This system and its expert team can, and has been, deployed for ARL research to provide precise ground truth of multiple moving and fixed objects, pedestrians and vehicles in support of pedestrian detection experiments. Working with GDRS, this system was used to support multiple data collections on sites as large as 400 x 200 m with accuracy of approximately 20 cm.


An UWB receiver, tags, deployment on mast, tags on hats, and the GDRS/CTA Westminster, MD test site.



4. NIST brings its unique measurement background to every aspect of measurement exercises, eliminating sources of ambiguity. For example, NIST employed specialized hardware at no cost to ARL to ensure the time stamps used by ANS and NIST computers could be synchronized in a recent data collection. Similarly, NIST devised and implemented an independent method to ensure initial heading of the test vehicle was repeatable, without reliance on the vehicle on-board navigation system, in a recent ARL evaluation of CTA systems.
5. NIST’s Intelligent Systems Division (ISD) has a long history and considerable expertise in intelligent control architectures and mobile robotics, providing valuable insight into the definition of appropriate metrics and suitable measurement methods. ISD designed the 4D/RCS Reference Model Architecture that was used by the ARL Demo III program and has been adopted by Boeing, SAIC, and GDRS for the FCS Autonomous Navigation System (ANS). ISD currently applies its expertise to measurement needs of ARL, DARPA, FHWA, DHS and others.
6.  NIST has multiple test bed vehicles available for use in ARL work. In addition to the ARL/NIST autonomous HMMWV, NIST has vehicles such as a Chrysler sedan, that is instrumented with various sensors and the same precise position and orientation system installed on the HMMWV. Position and orientation data is available at up to 200 Hz on these platforms. These platforms and others can be, and are, leveraged to assist in sensor data collection exercises. A test van, fitted to support field work, is also available and routinely used in support of such exercises as those for the ARL ANS and CTA events.
7. NIST has both a large (500+ acre) site and a smaller NIKE Annex site throughout which various tests and measurements may be developed and performed. Each has been precisely mapped via aerial survey (6” accuracy) and high resolution maps and orthophotos are available. Each is covered by an RTK differential GPS system, providing real-time location accuracies of 2-3 cm for vehicles and objects on the sites. Together, these sites provide an instrumented test environment, with a wide range of physical characteristics, including open spaces, paved roads with and without lane markings, wooded space, cluttered urban-like environments, hills, ponds, creeks and ditches, etc. All large, and most small, structures are included in the survey data.




The NIST NIKE Annex site and example vehicle and pedestrian paths used in recent ANS data collection at NIKE site.



8. An extensive simulation environment (MOAST) is available with various portions of the campus implemented.
9. NIST has established relationships with managers of test track facilities in the area. Two examples are the MD Fire/Rescue and Police Training Academy in Rockville, MD and the Maryland Police and Correctional Training Commissions Driver Training Facility in Sykesville, MD. The latter is equipped with high speed and urban roadway test tracks as well as open skid pads, and has been used by NIST at no cost.




Sykesville test tracks. High speed loops, urban and skidpads, intersection closeup  



10. The NIST facility is fenced and guarded, and a Mobility Laboratory, with its own secure access houses test vehicles.  Further, a laboratory has been specially fitted with security measures in support of ANS work.