Achieving integrated convoys: cargo unmanned ground vehicle development and experimentation

Noah Zych, David Silver, David Stager, Colin Green, Thomas Pilarski, Jacob Fischer, Noah Kuntz, D.m. Anderson, Albert Costa, Joseph Gannon, Joseph Lisee, Peter Rander, Michael K. Sergi-Curfman, Christopher J. Shaw, Daniel Tascione, Nicolas Vandapel, John BeckProceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2013 · DOI

TL;DR

This paper details the development and experimental validation of an autonomous system for integrating unmanned cargo vehicles into existing military logistics convoys. The system successfully demonstrated the potential for enhanced force protection and increased operational efficiency, with warfighters expressing confidence in controlling multiple unmanned assets.

What Problem Does This Paper Address?

The paper addresses the critical need to enhance the safety and efficiency of military logistics operations, particularly in environments prone to ambushes and improvised explosive device (IED) attacks. Traditional logistics convoys place warfighters at significant risk. The core problem is to develop and integrate autonomous capabilities into tactical wheeled vehicles (TWVs) that can seamlessly join manned convoys, providing a force protection advantage by creating standoff distance and acting as force multipliers by allowing a single operator to manage multiple unmanned assets. This aims to reduce risk and manpower requirements for resupply missions.

Methods

The Cargo UGV project involved developing a comprehensive autonomous system integrated onto two Oshkosh Medium Tactical Vehicle Replacements (MTVRs). Key methodological components include:

  • System Architecture: A modular design comprising vehicle control, perception, and motion planning software.
  • Vehicle Control: A "drive-by-wire" system with actuation, power control, diagnostics, and safety monitoring subsystems, utilizing CAN 2.0b and SAE J1939 protocols for communication.
  • Perception System: Utilizes LADAR and radar sensors for real-time environmental understanding. This includes:
  • Terrain Detection: Ray tracing and LADAR data for ground surface analysis, generating elevation, slope, and roughness maps.
  • Moving Obstacle Detection: Fuses radar and LADAR signals using a sigma point unscented Kalman filter to track and report moving objects.
  • Radar Evidence Grid: Fuses radar measurements for a single model of occupied and empty terrain, complementing LADAR.
  • Terrain Classification: Fuses LADAR and evidence grid data to assign semantic labels to terrain patches, differentiating drivable areas from hazards.
  • Perception Aggregation and Costing: Aggregates terrain properties into a mobility cost map for the motion planner.
  • Map Registration and RNDF Alignment: Uses a map-based positioning approach to correct GPS drift and align with road networks.
  • Motion Planning: Emulates human driving and convoy behavior, generating real-time trajectories and commands. This includes:
  • Local Planner: Generates obstacle-free trajectories offset from a global path.
  • Speed Planning: Generates speed profiles considering acceleration limits, slope, and dispersion.
  • Path Follower: Corrects tracking errors and generates vehicle commands.
  • Moving Obstacle Behaviors: Integrates predicted paths of moving obstacles into the cost map.
  • Dispersion Behaviors: Manages vehicle spacing within the convoy.
  • Operator Control Unit (OCU): A custom human-machine interface for mission command and control, featuring an overhead map, live video feeds, and mission planning tools.
  • Simulation Environment: A realistic convoy simulation environment was created for rapid system development and testing.
  • Operator Training Program: A 3-day curriculum including classroom lessons, simulations, and hands-on practice was developed for novice operators.

Datasets

Not explicitly detailed in terms of specific datasets used for training machine learning components. However, the paper mentions:

  • Perception logs: Collected and used to construct a simulation environment for offline motion planning development.
  • Human-labeled data sets: Used for training the terrain classification system.
  • Vehicle data: Collected to correlate radar measurements to the presence or absence of geometric objects.

The paper does not introduce new benchmark datasets but rather focuses on the development and evaluation of a system using existing data and simulated environments.

Key Insights and Ideas

  • Seamless Integration is Key: The success of autonomous vehicles in logistics convoys hinges on their ability to be "unobtrusively integrated," meaning they must behave and appear similar to manned vehicles.
  • Multi-Modal Sensing for Robustness: Combining LADAR and radar provides crucial redundancy and robustness, especially in challenging conditions like dust or poor visibility.
  • Human-in-the-Loop for Complex Scenarios: While aiming for autonomy, the system design allows for operator intervention and control (tele-operation, "go-zone" commands) for situations where full autonomy is not feasible or safe.
  • Simulation as a Development Accelerator: A realistic convoy simulation environment significantly sped up development and validation by allowing for rapid iteration and testing of software modules.
  • Effective Training is Crucial: A well-designed training program is essential for warfighters to gain confidence and proficiency in operating and supervising unmanned vehicles in complex convoy operations.

Key Results

  • Successful Autonomous Operation: The Cargo UGV system was successfully developed and demonstrated autonomous operation in complex, operationally representative environments.
  • Force Protection Potential: The system validated the hypothesis that unmanned tactical wheeled vehicles provide a force protection advantage by creating standoff distance.
  • Force Multiplier Capability: The system demonstrated that a single operator can comfortably control multiple unmanned assets (3-5 projected).
  • Extensive Field Testing: Over 650 km of autonomous driving was accomplished during a one-week test, with an average speed of over 28 kph.
  • Safe Operation: Demonstrated safe operation at sustained speeds over 55 kph and successful navigation through challenging terrain, including GPS denial scenarios.
  • Warfighter Confidence: Warfighters surveyed expressed high confidence in the system's capabilities, exceeding original expectations.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges the challenge of delivering a vehicle capable of operating dependably in "often-uncertain environments relevant to logistics units," which include interactions with live friendly forces, neutral local populace, and other manned/unmanned vehicles.
  • Wireless communication dropouts in a convoy remain a significant challenge, requiring sophisticated reaction behaviors.
  • The system's performance in extreme weather conditions beyond those tested is not fully detailed.
  • Future Work:
    • The authors project that the integration of unmanned vehicles into operational environments will occur in the near future, implying continued development and deployment.
  • The paper suggests the need for further evolution of concepts of operation (CONOPs) and tactics, techniques, and procedures (TTPs) for the successful employment of large UGVs in future military conflicts.
  • The development of more advanced machine learning techniques and toolchains for producing and validating results is mentioned as an ongoing effort.

Summary updated

Abstract

The Cargo UGV project was initiated in 2010 with the aim of developing and experimenting with advanced autonomous vehicles capable of being integrated unobtrusively into manned logistics convoys. The intent was to validate two hypotheses in complex, operationally representative environments: first, that unmanned tactical wheeled vehicles provide a force protection advantage by creating standoff distance to warfighters during ambushes or improvised explosive device attacks; and second, that these UGVs serve as force multipliers by enabling a single operator to control multiple unmanned assets. To assess whether current state-of-the-art autonomous vehicle technology was sufficiently capable to permit resupply missions to be executed with decreased risk and reduced manpower, and to assess the effect of UGVs on customary convoy tactics, the Marine Corps Warfighting Laboratory and the Joint Ground Robotics Enterprise sponsored Oshkosh Defense and the National Robotics Engineering Center to equip two standard Marine Corps cargo trucks for autonomous operation. This paper details the system architecture, hardware implementation, and software modules developed to meet the vehicle control, perception, and planner requirements compelled by this application. Additionally, the design of a custom human machine interface and an accompanying training program are described, as is the creation of a realistic convoy simulation environment for rapid system development. Finally, results are conveyed from a warfighter experiment in which the effectiveness of the training program for novice operators was assessed, and the impact of the UGVs on convoy operations was observed in a variety of scenarios via direct comparison to a fully manned convoy.
BibTeX
@inproceedings{zych2013achieving,
  author = {Noah Zych and David Silver and David Stager and Colin Green and Thomas Pilarski and Jacob Fischer and Noah Kuntz and D.M. Anderson and Albert Costa and Joseph Gannon and Joseph Lisee and Peter Rander and Michael K. Sergi-Curfman and Christopher J. Shaw and Daniel Tascione and Nicolas Vandapel and John Beck},
  title = {{Achieving integrated convoys: cargo unmanned ground vehicle development and experimentation}},
  booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
  year = {2013},
  doi = {10.1117/12.2015586},
}
Figure