3D SLAM: Simultaneous Localization and Mapping in 3D Scott Robbins, Principal Investigator Problems:
In the cluttered environments of urban and interior settings, narrow field of view cameras and poor localization can lead to a loss of situation awareness. In these environments it becomes very difficult to judge the relative positions of objects seen recently with objects currently observed. While 2D video mosaics can address this problem for higher altitude UAVs, motion parallax from operations at very low or ground levels break this technique. For these 3D environments we will investigate 3D simultaneous localization and mapping to provide improved operator situation awareness.
Objectives:
We will investigate techniques for 3D sensor exploitation through real-time SLAM suitable for use on small UAVs and UGVs operating in urban canyon and interior environments. By providing both human and machine operators with a synthesized third person "3D mosaic" and alternate localization to GPS, we will demonstrate improvements to situational awareness in these critical areas.
Activities:
We will identify and test various approaches for real-time 3D SLAM suitable for application to small military UGVs and UAVs. Localization approaches will be characterized in terms of precision and drift from ground truth. Mapping approaches will be characterized in terms of fidelity to the operational environment and mission requirements.
Impact:
Operations in urban canyons and interior environments are among the most dangerous activities in which our sponsors are currently engaged. The 3D sensing and exploitation provided by simultaneous localization and mapping will greatly enhance the utility of small unmanned vehicles to provide situational awareness in settings not well served by currently deployed systems. This research will provide a set of demonstrated technologies for 3D localization and mapping suitable for integration into military small UGVs and UAVs operating in urban and interior environments.
Approved for Public Release: 08-0447 Presentation [PDF]
Advanced Perception for Unmanned Ground Vehicles Bob Grabowski, Principal Investigator Problems:
Perception is essential for competent ground robots. Operational expectations for existing robots far exceed current technology as evidenced by recent events like the Urban Grand Challenge. This, in large part, is due to the limitations of existing sensor technology. Robots derive perception of the world directly through sensors. As the mission space becomes more complex, sensor technology must keep pace.
Objectives:
Demonstrate improved perception capability on an unmanned ground vehicle (UGV). Provide a framework for defining the necessary sensing capabilities for unmanned ground vehicles. Demonstrate how to enhance vehicle perception by combining multiple sensors to overcome the weakness of an individual technology. Investigate novel techniques such as sensor fusion, sensor cueing, sensor steering, and mosaicing.
Activities:
Identify capabilities of existing sensor technology with respect to four key areas of UGV perception: maneuvering, navigation, safety, and awareness. Characterize the strengths and weaknesses of each approach. Develop new methods for combining and fusing information from multiple sensors to overcome identified limitations and gaps. Test and verify new approaches in a representative urban environment on a real autonomous vehicle.
Impact:
This research will shape how to approach future development of sensing for unmanned ground vehicles with respect to both the vehicle and humans that must operate around it. By developing and testing on an existing autonomous vehicle, this research will also yield a vehicle capable of conducting more complex missions in realistic, operationally relevant environments.
Approved for Public Release: 08-0293 Presentation [PDF]
Disposable Walking Robots Adam McLeod, Principal Investigator Problems:
Getting somewhere is a common precondition to doing something. However, designing a robot to get somewhere typically restricts what it can do. Design choices enter into a network of tradeoffs between where one needs to go, what one needs to do, and how much one is willing to spend to get there and do it.
Objectives:
We are investigating methods to design agile robots that can traverse difficult terrain, accomplish a task at its destination, and can be pervasively deployed easily. Research will focus on using a human operator to train a walking robot to ambulate over difficult terrain.
Activities:
Two major experiments will be conducted using a simple exoskeleton for data collection and a semi-autonomous walking machine. The first experiment will use the simple exoskeleton to collect kinematic data for algorithm prototyping and platform development. The second will attempt to train the walking machine to ambulate over varied, complex terrain.
Impact:
This project will design methodologies for walking robots that will increase the speed and efficiency when developing robotic solutions for Explosive Ordinance Disposal (EOD), reconnaissance, or any other task requiring pervasive deployment.
Approved for Public Release: 08-0659
Lightweight Beacon System for UAS and Other Aviation Applications Robert Strain, Principal Investigator Problems:
The emergence of unmanned and sport aviation class aircraft has introduced an increased safety risk. Current transponder and ADS-B units require aircraft electrical systems, are relatively heavy and expensive, and won't likely be required in uncontrolled airspace. ADS-B technology offers an opportunity to address this risk, but it must be affordable and usable on these special use aircraft.
Objectives:
We will demonstrate a small, low-powered beacon transceiver suitable for less-maneuverable or special use aircraft operating in national airspace to improve visibility to proximate aircraft. The beacon makes use of the Universal Access Transceiver (UAT) waveform and complies with ADS-B performance requirements. A modular architecture enables it to be either stand-alone or integrated with other electronics or sensors.
Activities:
We have developed a small, self-contained, battery-operated transmitter unit, and tested it on a small commercial Unmanned Aircraft System (UAS). Transmissions can be received at ranges in excess of 20 miles, and battery power lasts 14 hours. In FY08 receiver functionality will be added. The technology will be made available for licensing to commercial companies.
Impact:
Availability of an affordable beacon for all manned and unmanned aircraft has the potential to enhance safety by improving overall pilot situational awareness. The availability of such technology could also be beneficial in many national security, public safety, and disaster relief scenarios by improving operational effectiveness and reducing response times.
Approved for Public Release: 08-0022 Presentation [PDF]
Sense and Avoid for Small Unmanned Aircraft David Maroney, Principal Investigator Problems:
This research will address the question, "Can small autonomous aircraft reliably detect conflict and avoid collision (by reacting predictably) with objects in their path, both stationary and moving, that do not announce their position?" Many research efforts focus on one component or technology of this "sense-detect-avoid" question, but the end-to-end system has not been widely addressed.
Objectives:
The project will investigate sensing technologies, obstacle detection algorithms, and avoidance methods to assemble unique combinations for comprehensively addressing the research question. We will analyze and map the breadth of sense-detect-avoid concepts and technology, scope appropriate alternatives for small UAS, and understand their viability by building, testing, and flying selected combinations through a progressive set of laboratory and field tests.
Activities:
Previously we framed the spectrum of sensors, and selected four for experimentation -- video, laser range finder, ultra-wide band radar, and acoustic array. We have developed autonomous airframes and have performed sense and detection tests with two sensors on stationary and moving targets. This year, we will complete testing and data collection for these sensors and develop detect-and-avoid algorithms.
Impact:
Direct impacts will include influencing collision avoidance capabilities of small UAS, improving survival of UAS and unintentional targets, and informing policy decisions on certification and regulation. We developed effective cross-MITRE collaboration between robotics and aviation areas, and contributed to industry recognition of MITRE involvement. We published three papers at AIAA and IEEE conferences, and have presented in a number of technical forums.
Approved for Public Release: 08-0014 Presentation [PDF]
UGV Leader-Follower Bob Grabowski, Principal Investigator Problems:
Robot capabilities are evolving beyond simple actions that extend the presence and capability of the human operator to agents that can fulfill more involved tasks such as following a route, tracking a target and patrolling a fence.
Objectives:
These robots will coordinate to provide physical presence and overwatch of secured areas while avoiding hazards. The robots will work as a coordinated team to insure area coverage, resource scheduling and, leader follower of robot groups. The investigation will include how to monitor a given area, insure coverage, coordinate assets and relay and disseminate information back to a central user.
Activities:
We will demonstrate these capabilities in a mock force protection mission scenario involving the Meteor, Mobile Robot Lab and one or more Centaur vehicles in an outdoor parking lot. The robots will follow a leader to a remote area and conduct surveillance and reconnaissance while reporting status and anomalies back to the command post.
Impact:
This research will provide valuable risk assessment to the FCS community in how to task and monitor UGVs from a remote station. It will examine the need and roles of the human in the loop with respect to controlling multiple unmanned ground vehicles and investigate how much can be unburdened through autonomy.
Approved for Public Release: 08-0400
Vision for Estimating Terrain Traversability Jeff Colombe, Principal Investigator Problems:
Current sensors on UGVs are limited to near-field sensing (<50m), with no information about the material properties of terrain. As a result, they have limited capability to judge where and how they should drive in order to reach a desired location.
Objectives:
We will exploit the dependencies between terrain appearance and terrain mechanics to interpret the traversability of terrain visible at long range using video sensors. Our analysis will include preprocessing of stereo and motion information in video during vehicle navigation to reveal useful visual features, followed by statistical regression to identify what mechanical properties of terrain are evidenced by visual information.
Activities:
We will mount sensors on MITRE UGVs, passively record data during other vehicle activities, then do intensive image processing and statistical analysis of data. The results will include 3D environment maps that predict various mechanical properties of terrain, including quality of control of traction, steering, and braking, vehicle attitude, and motion derivatives.
Impact:
This work will contribute to UGV duty cycle transitioning from teleoperation to 100% autonomous behavior. Sponsors that will benefit include Army Future Combat Systems, and other consumers of image processing technology in the national security space.
Approved for Public Release: 07-1551
Last Updated:05/05/2008 | ^TOP | 
