According to a recent report (1) between 50 and 80 countries either already use robotic systems in defence applications or are in the process of acquiring or developing such technology. The majority of these systems comprise unmanned vehicles for air, land and sea operation: unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGVs), and unmanned underwater vehicles (UUVs). The rapid rate of adoption of unmanned systems in the US military is illustrated by the fact that in 2003, when the US invaded Iraq, it had only a handful of UAVs and no UGVs; by 2010 those numbers had grown to in excess of 7000 UAVs and 12 000 UGVs (2). By 2009 more US Air Force pilots were being trained to fly unmanned aerial systems from ground operations centres than pilots to fly fighter or bomber aircraft(3).
A typical scenario
The US Army introduced the Robotics Rodeo in 2009 as a venue for prominent robotic technologists to demonstrate how unmanned systems can support current and future military, disaster relief and humanitarian support efforts.
At the 2010 Robotics Rodeo a typical scenario was enacted where two UAVs (quarter-scale Piper Cubs) were used to locate a target in an urban setting and guide a UGV (a highly modified Porsche Cayenne operating autonomously) to the target: the onboard computer systems on each of the three robots combining to form a collaborative autonomous system.
As more robots are introduced into and more humans are removed from scenarios where there is limited or unreliable distance communications the degrees of autonomy of the robots need to increase in terms of energy and decision making. Technologies need to be developed to meet the resulting challenges of:
* Autonomous decision making.
* Software verification and validation.
* Autonomous communication for interoperation and avoidance.
* Reliable local communication infrastructure between vehicles.
And commercial interest is necessary to develop these technologies – to ensure that there are sufficient research and development resources, manufacturers and engineers developing the skills to meet these challenges and produce the vehicles and systems.
Autonomous decision making
Essential prerequisites for a robotic controller to make decisions autonomously include:
* Accurate object recognition.
* Situational understanding.
Lora Weiss (4), a lab chief scientist at the Georgia Tech Research Institute, recently wrote that despite being equipped with high numbers of sophisticated sensors, “[M]ost UAVs cannot distinguish a sleeping dog from a bush, even at high noon.”
Even once the recognition problem (accurate visual cognition) is overcome, Weiss points out that a key requirement for armed unmanned robots is the ability to correctly interpret situations, even when faced with incomplete or conflicting data, before deciding to their weaponry. “If a robo-sentry armed with a semiautomatic rifle detects someone running from a store, how can it know whether that person has just robbed the store or is simply sprinting to catch a bus? Does it fire its weapon based on what it thinks is happening?”
Software verification and validation
The decision making software that is used for autonomous vehicle control has to be able to cope with unforeseen situations, and incomplete or conflicting data. Logic falls into the realm of artificial intelligence and fuzzy logic where testing by observation is well-nigh impossible. In addition software needs to be verifiably free of any malware, hidden triggers, backdoors, hardcoded passwords and other security risks. Automated model-based code generation, simulation and automated testing are all areas in which significant developments are under way.
In the early phases of autonomous robot development, little thought was given to inter-robot communication standards. It was assumed that robots would work co-operatively with human handlers or partners. But as autonomous robots have been deployed in increasing numbers, and as the realm of micro-robots and swarm robotics have developed, the need for robots from different vendors to intercommunicate has become more important: both for collaborative tasks and for avoidance.
Historically there are two main communication protocols that have been adopted for unmanned vehicle intercommunication:
* The Joint Architecture for Unmanned Systems (JAUS).
* NATO Standardisation Agreement (STANAG) 4586.
Both of these protocols are message-based. The primary focus of JAUS is UGVs while that of STANAG 4586 is UAVs. But as seen in the typical scenario above there are situations where UAVs and UGVs need to work together as a collaborative system. JAUS offers more by way of command and control functionality while STANAG 4586 offers more in terms of data handling from (UAV) payloads. Both of these standards incorporate a MIL-STD-1760 interface for weapon control and both have in the last few years implemented a Service Oriented Architecture (SOA).
More recently the Data Distribution Service (DDS ) standard for real-time and embedded systems managed by the Object Management Group (OMG) has been adopted as a comprehensive middleware solution for ‘publish-subscribe’ data-centric integration in mission critical and safety systems. One of the key benefits of the OMG DDS is the ability to specify quality of service (QoS) parameters per data stream. For instance a contract between a data publisher and a data subscriber can specify parameters such as update frequency, level of reliability, delivery order, data durability and prioritisation.
Because DDS is a peer-to-peer solution it is more fault-tolerant than systems that route through brokers or bridges.
DDS has been applied in civilian and military mission critical applications such as (5):
* VW Driver Assistance & Integrated Safety system.
* NASA robotics.
* CAE Sim XXI flight simulator.
* ATLANTIDA consortiums European air traffic management system.
* The US Army Joint Battle Command Platform (JBC-P Blue Force Tracker).
* US Navy Aegis Open Architecture (AOA).
* US Navy Zumwalt DDG 1000 destroyer.
* RUAG BASE TEN RoboScout satellite-based remote-control UGV.
Reliable local communication and mapping
In many situations autonomous robots are required to operate in unknown landscapes, such as inside buildings where floor plans are unavailable, normal positioning technology like GPS will not operate and communication paths are attenuated by obstacles.
For a robot or a swarm of robots to function effectively in such an environment, as for instance in a search and rescue mission, it needs to know the topology in which it is working, starting from the unknown position in terms of reference coordinates in which it finds itself. Additionally individual robots need to be able to communicate with each other and establish/maintain a line of communication to the outside world.
Commercial motes from companies like JLH Labs and Libelium allow micro-robots to establish their own short-range mesh network as they explore and map regions.
The robots then use a technique known as simultaneous localisation and mapping (SLAM) to build a map of the unknown environment while at the same time keeping track of their current location within the frame of reference which they are constructing.
Crowd-source a military robot?
Novel approaches that DARPA has applied to design and testing include crowd-sourcing and game-based platforms for simulation. In February 2011 DARPA offered a $10 000 prize in the Experimental Crowd-derived Combat-support Vehicle (XC2V) Design Challenge – experts, novices and the curious were invited to provide input to the final design of the vehicle or vote on the best submission.
In March 2011 the Agency announced the public availability of its autonomous robotics manipulation (ARM) simulator. Participants can download this simulator, use it to develop code that can subsequently be uploaded to the ARM team and then watch in real-time over the web as the team runs their code on an actual ARM robot. DARPA claims that the development of complex software systems has benefitted significantly from the ability to leverage crowd-sourced innovation in the form of open source code development.
In April 2011 the Agency announced the integration of its anti-submarine warfare (ASW) continuous trail unmanned vessel (ACTUV) Tactics Simulator into the Dangerous Waters game, which it has made available for public download. This software was written to simulate actual evasion techniques used by submarines, challenging each player to track them successfully. Players earn points based on mission success and can rank their effectiveness on DARPA’s leader board web page. As players complete each scenario in the simulation, they can submit their tracking tactics to DARPA for analysis. DARPA will select the best tactics and build them into the ACTUV prototype.
Autonomous robot: A robot that can perform desired tasks in unstructured environments without continuous human guidance (6)
DARPA: (US) Defense Advanced Research Projects Agency.
UAV: Unmanned aerial vehicle.
UGV: Unmanned ground vehicle.
UUV: Unmanned underwater vehicle.
Reading, viewing and resources
JLH Labs: Micro-wireless devices; http://www.jlhlabs.com/
Libelium: Micro-wireless devices; http://www.libelium.com/
Georgia Robotics and InTelligent Systems Lab: Video of UGVs and UAVs working in collaboration; http://tinyurl.com/3ulko6t
1. ABI Research: Defense Robots: UAVs, UGVs, UUVs and task robots for military applications; http://tinyurl.com/3en42kh
2. Brookings Institution: The unmanned mission; http://tinyurl.com/3kfpkze
3. The Washington Post: Fine Print: Air Force to train more remote than actual pilots; http://tinyurl.com/me6hld
4. IEEE: Spectrum: August 2011: Autonomous robots in the fog of war; http://tinyurl.com/3gb3983
5. UCS Architecture: Executive summary 2011: The data distribution service reducing cost through agile integration; http://tinyurl.com/3oytl8t
6. Wikipedia: Autonomous robot; http://en.wikipedia.org/wiki