A key component in the emerging localization and mapping paradigm is an appearance-based place recognition
algorithm that detects when a place has been revisited. This algorithm can run in the background at a low
frame rate and be used to signal a global geometric mapping algorithm when a loop is detected. An optimization
technique can then be used to correct the map by 'closing the loop'. This allows an autonomous unmanned ground
vehicle to improve localization and map accuracy and successfully navigate large environments. Image-based
place recognition techniques lack robustness to sensor orientation and varying lighting conditions. Additionally,
the quality of range estimates from monocular or stereo imagery can decrease the loop closure accuracy. Here,
we present a lidar-based place recognition system that is robust to these challenges. This probabilistic framework
learns a generative model of place appearance and determines whether a new observation comes from a new or
previously seen place. Highly descriptive features called the Variable Dimensional Local Shape Descriptors are
extracted from lidar range data to encode environment features. The range data processing has been implemented
on a graphics processing unit to optimize performance. The system runs in real-time on a military research
vehicle equipped with a highly accurate, 360 degree field of view lidar and can detect loops regardless of the
sensor orientation. Promising experimental results are presented for both rural and urban scenes in large outdoor
The Multi-Agent Tactical Sentry Unmanned Ground Vehicle, developed at Defence R&D Canada - Suffield, has
been in service with the Canadian Forces for five years. This tele-operated wheeled vehicle provides a capability
for point detection of chemical, biological, radiological, and nuclear agents. Based on user experience, it is
obvious that a manipulator capability would greatly enhance the vehicle's utility and increase its mobility in
urban terrain. This paper details technical components of this development, and describes a number of trials
undertaken to perform tasks with a manipulator arm such as picking up objects, opening vehicle and building
doors, recording video, and creating 3D models of the environment. The lessons learned from these trials will
guide further development of the technology.
Unmanned vehicles (UxV) operate in numerous environments, with air, ground and marine representing the
majority of the implementations. All unmanned vehicles, when traversing unknown space, have similar requirements.
They must sense their environment, create a world representation, and, then plan a path that safely
avoids obstacles and hazards. Traditionally, each unmanned vehicle class used environment specific assumptions
to create a unique world representation that was tailored to it operating environment. Thus, an unmanned aerial
vehicle (UAV) used the simplest possible world representation, where all space above the ground plane was free
of obstacles. Conversely, an unmanned ground vehicle (UGV) required a world representation that was suitable
to its complex and unstructured environment.
Such a clear cut differentiation between UAV and UGV environments is no longer valid as UAVs have migrated
down to elevations where terrestrial structures are located. Thus, the operating environment for a low flying
UAV contains similarities to the environments experienced by UGVs. As a result, the world representation
techniques and algorithms developed for UGVs are now applicable to UAVs, since low flying UAVs must sense
and represent its world in order to avoid obstacles.
Defence R&D Canada (DRDC) conducts research and development in both the UGV and UAV fields. Researchers
have developed a platform neutral world representation, based upon a uniform 21/2-D elevation grid,
that is applicable to many UxV classes, including aerial and ground vehicles. This paper describes DRDC's
generic world representation, known as the Global Terrain map, and provides an example of unmanned ground
vehicle implementation, along with details of it applicability to aerial vehicles.
The Autonomous Intelligent Systems Section at Defence R&D Canada - Suffield envisions autonomous systems
contributing to decisive operations in the urban battle space. In this vision, teams of unmanned ground, air, and
marine vehicles, and unattended ground sensors will gather and coordinate information, formulate plans, and
complete tasks. The mobility requirement for ground-based mobile systems operating in urban settings must
increase significantly if robotic technology is to augment human efforts in military relevant roles and environments.
In order to achieve its objective, the Autonomous Intelligent Systems Section is pursuing research that
explores the use of intelligent mobility algorithms designed to improve robot mobility. Intelligent mobility uses
sensing and perception, control, and learning algorithms to extract measured variables from the world, control
vehicle dynamics, and learn by experience. These algorithms seek to exploit available world representations of
the environment and the inherent dexterity of the robot to allow the vehicle to interact with its surroundings
and produce locomotion in complex terrain. However, a disconnect exists between the current state-of-the-art
in perception systems and the information required for novel platforms to interact with their environment to
improve mobility in complex terrain. The primary focus of the paper is to present the research tools, topics, and
plans to address this gap in perception and control research. This research will create effective intelligence to
improve the mobility of ground-based mobile systems operating in urban settings to assist the Canadian Forces
in their future urban operations.
In support of Canadian Forces transformation, Defence R&D Canada (DRDC) has established an ongoing program to develop machine intelligence for semi-autonomous vehicles and systems. Focussing on mine clearance and remote scouting for over a decade, DRDC Suffield has developed numerous UGVs controlled remotely over point-to-point radio links. Though this strategy removes personnel from potential danger, DRDC recognized that human factors and communications bandwidth limit teleoperation and that only networked, autonomous unmanned systems can conserve these valuable resources. This paper describes the outcome of the first autonomy project, Autonomous Land Systems (ALS), designed to demonstrate basic autonomous multivehicle land capabilities.
In 2002 Defence R&D Canada changed research direction from pure tele-operated land vehicles to general autonomy for land, air, and sea craft. The unique constraints of the military environment coupled with the complexity of autonomous systems drove DRDC to carefully plan a research and development infrastructure that would provide state of the art tools without restricting research scope. DRDC's long term objectives for its autonomy program address disparate unmanned ground vehicle (UGV), unattended ground sensor (UGS), air (UAV), and subsea and surface (UUV and USV) vehicles operating together with minimal human oversight. Individually, these systems will range in complexity from simple reconnaissance mini-UAVs streaming video to sophisticated autonomous combat UGVs exploiting embedded and remote sensing. Together, these systems can provide low risk, long endurance, battlefield services assuming they can communicate and cooperate with manned and unmanned systems. A key enabling technology for this new research is a software architecture capable of meeting both DRDC's current and future requirements. DRDC built upon recent advances in the computing science field while developing its software architecture know as the Architecture for Autonomy (AFA). Although a well established practice in computing science, frameworks have only recently entered common use by unmanned vehicles. For industry and government, the complexity, cost, and time to re-implement stable systems often exceeds the perceived benefits of adopting a modern software infrastructure. Thus, most persevere with legacy software, adapting and modifying software when and wherever possible or necessary -- adopting strategic software frameworks only when no justifiable legacy exists. Conversely, academic programs with short one or two year projects frequently exploit strategic software frameworks but with little enduring impact. The open-source movement radically changes this picture. Academic frameworks, open to public scrutiny and modification, now rival commercial frameworks in both quality and economic impact. Further, industry now realizes that open source frameworks can reduce cost and risk of systems engineering. This paper describes the Architecture for Autonomy implemented by DRDC and how this architecture meets DRDC's current needs. It also presents an argument for why this architecture should also satisfy DRDC's future requirements as well.
In order for an Unmanned Ground Vehicle (UGV) to operate effectively it must be able to perceive its environment in an accurate, robust and effective manner. This is done by creating a world representation which encompasses all the perceptual information necessary for the UGV to understand its surroundings. These perceptual needs are a function of the robots mobility characteristics, the complexity of the environment in which it operates, and the mission with which the UGV has been tasked. Most perceptual systems are designed with predefined vehicle, environmental, and mission complexity in mind. This can lead the robot to fail when it encounters a situation which it was not designed for since its internal representation is insufficient for effective navigation. This paper presents a research framework currently being investigated by Defence R&D Canada (DRDC), which will ultimately relieve robotic vehicles of this problem by allowing the UGV to recognize representational deficiencies, and change its perceptual strategy to alleviate these deficiencies. This will allow the UGV to move in and out of a wide variety of environments, such as outdoor rural to indoor urban, at run time without reprogramming. We present sensor and perception work currently being done and outline our research in this area for the future.
Proc. SPIE. 5422, Unmanned Ground Vehicle Technology VI
KEYWORDS: Defense and security, Unmanned aerial vehicles, Sensors, Robotics, Control systems, Artificial intelligence, Intelligence systems, Algorithm development, Systems modeling, Decision support systems
The Defence Research and Development Canada's (DRDC has been given strategic direction to pursue research to increase the independence and effectiveness of military vehicles and systems. This has led to the creation of the Autonomous Intelligent Systems (AIS) prgram and is notionally divide into air, land and marine vehicle systems as well as command, control and decision support systems. This paper presents an overarching description of AIS research issues, challenges and directions as well as a nominal path that vehicle intelligence will take. The AIS program requires a very close coordination between research and implementation on real vehicles. This paper briefly discusses the symbiotic relationship between intelligence algorithms and implementation mechanisms. Also presented are representative work from two vehicle specific research program programs. Work from the Autonomous Air Systems program discusses the development of effective cooperate control for multiple air vehicle. The Autonomous Land Systems program discusses its developments in platform and ground vehicle intelligence.
The mobility requirement for Unmanned Ground Vehicles (UGVs) is expected to increase significantly as the number of conflicts shift from open terrain operations to the increased complexity of urban settings. In preparation for this role Defence R&D Canada-Suffield is exploring novel mobility platforms utilizing intelligent mobility algorithms that will each contribute to improved UGV mobility. The design of a mobility platform significantly influences its ability to maneuver in the world. Highly configurable and mobile platforms are typically best suited for unstructured terrain. Intelligent mobility algorithms seek to exploit the inherent dexterity of the platform and available world representation of the environment to allow the vehicle to engage extremely irregular and cluttered environments. As a result, the capabilities of vehicles designed with novel platforms utilizing intelligent mobility algorithms will outperform larger vehicles without these capabilities. However, there exist many challenges in the development of UGV systems to satisfy the increased mobility requirement for future military operations. This paper discusses a research methodology proposed to overcome these challenges, which primarily involves the definition and development of novel mobility platforms for intelligent mobility research. It addresses intelligent mobility algorithms and the incorporation of world representation and perception research in the creation of necessary synergistic systems. In addition, it presents an overview of the novel mobility platforms and research activities at Defence R&D Canada-Suffield aimed at advancing UGV mobility capabilities in difficult and relevant military environments.