SubscribeThe Vehicle Subsystems and Robotics Group at QinetiQ have been at the forefront of research and development into unmanned ground vehicles in recent years.
In a new era of 21st century combat, the team is working on the development of a new breed of artificially intelligent vehicles.
The vision of a battlefield populated entirely by robots engaged in armed combat may seem more akin to Bladerunner and The Terminator than real warfare. Though the development of such sophisticated robots may be many years away, the prospect of a digitised battlefield where unmanned land vehicles can see and think for themselves may not be as far off as we imagine.
Research into Unmanned Ground Vehicles (UGVs) began more than sixty years ago during World War Two. UGVs have been playing an active role in modern warfare for more than thirty years and fulfil a broad range of applications in activities such as route opening, obstacle breaching and mine clearance both during and immediately after combat and in peace time operations.
“These tasks can be extremely hazardous and the use of remotely controlled vehicles can alleviate or, at least, reduce the danger for army personnel in many of these situations”, explains Bill Warren, Group Project Manager at QinetiQ.
The group has played a vital role in research for the MOD in recent years, focusing on teleoperation of large engineer equipment such as the 18 tonne Combat Engineer Tractor (CET) and the AVRE (Armoured Vehicle Royal Engineers) used in routine, but essential engineering tasks. Indeed much of this work was carried out as part of an urgent operational requirement for the MOD to aid mine clearance in Bosnia and Kosovo.
The key benefit of teleoperation is safety –the vehicle is operated from several kilometres away using video images transmitted from the vehicle to the controller via microwave links. Man-in-the-loop control of the vehicle is maintained via military or commercial radio communications but, as the operators are far removed from the vehicle during hazardous operations, the risk to human life is greatly reduced.
Extending the battlefield
This is not the only benefit of UGVs. Removing the human risk factor also opens up the possibility for much bolder concepts of operation. The use of robotics can also bring about higher performance levels for lengthy and repetitive tasks where humans may suffer fatigue, boredom or stress, as well as releasing personnel for more technically demanding activities.
“Winning the war on the battlefield is all about improving performance and so maintaining the technological edge is absolutely crucial ”, says Bill. Over the past five years, the QinetiQ team has also been closely involved in the development of teleoperation systems for the completion of complex, manipulative tasks in both the construction and clearance of obstacles.
More specifically, this work looked at the effectiveness of teleoperation of the CET for the completion of digging tasks that require the use of a mechanised bucket for earth removal.
“Our tests showed that tasks undertaken under remote control took around 10 –25 per cent longer than when attempted manually, but the major advantage is that they were conducted in a safe, reliable and robust manner, which were the key concerns for our MOD customers”, explains Neil Heyes, Technical Leader.
The system developed by QinetiQ can also be fitted to a range of existing or commercial vehicles, which can be switched back to manual operation whenever required. “Operators seem to adapt fairly quickly to teleoperation and generally only require a few hours of training”, says Neil.
There were some disadvantages, however. The main problem experienced by users was a lack of depth perception. Operators using mono-vision systems found it difficult to get a precise sense of what was happening to the vehicle and its surroundings.
Perceptive solutions
The main drawback of current classical teleoperation kits is a lack of multiple sensory information being relayed from vehicle to operator. This problem becomes apparent when operators are asked to perform even relatively simple tasks, using visual information only, and they experience difficulties.
With this in mind, the team investigated the integration of force feedback (via the user ’s joystick) using COTS technology designed for games systems, and on-board audio microphones linked to the operator ’s headsets. The idea was to increase the user ’s spacial awareness to enable them to control the vehicle more effectively.
Another problem with teleoperation is that the video images are transmitted via a microwave link that relies on line-of-sight. For obvious reasons, this can pose serious limitations in unknown territory. An example of this problem can be seen in television coverage of Formula One motor racing. Whenever the racing car passes behind an obstacle, such as a tunnel, the picture from the on board camera breaks up as the line-of-sight contact is lost.
Similarly, in unstructured outdoor environments, the operator risks losing visual communications, for example, when the vehicle drives behind a hill or a group of trees. If this happens, the operator is no longer in safe control of the vehicle.
To address this effect, the team started to look at alternative methods of visual data transmission. Real-time acquisition of data is an essential requirement in vehicle control, which ruled out techniques like video compression.
“We started to think through the problem in a more circumspect manner and looked at what information was needed and what could be eliminated”, explains Neil.
The team ’s theory was that, if the amount of data could be limited, useful data could be transmitted over a lower bandwidth and, instead of using video pictures, the scene could be represented via computer generated 3D graphics in the virtual world. These techniques are well developed in QinetiQ and can be applied to many different applications.
The team devised a system that works in a simple, yet effective way. Terrain data gathered from reconnaissance activities is used to create a map showing significant points of interest –such as ditches and obstacles.
As the vehicle navigates the terrain, positional data is transmitted to the control station via low bandwidth UHF radio that is used in turn to update the graphical image. The operator who is controlling the vehicle uses this information to adjust speed, direction and tasking.
Using virtual reality helps to overcome the depth perception problem encountered when viewing images that are transmitted from fixed cameras. Another advantage is that the scene can also be viewed from any angle during data update.
Unlike classic video systems, graphic visualisation is not dependent on ambient light, which means that operators can carry out tasks in adverse weather conditions (e.g. fog and snow) or even at night.
The digitised battlefield
Until now, research has been driven by the requirement for systems that retain man-in-the-loop command –such as teleoperation and virtual vision systems. However, the MOD is now looking to the development of a new generation of semi-autonomous and autonomous vehicles.
The Robotics team has recently begun an 18-month programme of research looking at aspects of artificial intelligence for autonomous navigation. This could lead to the development of a new breed of vehicles that have elements that can see and think for themselves.
In the future, we could see vehicles navigating across a battlefield in a tactical and intelligent manner, avoiding obstacles and enemy fire with minimal human input. These vehicles could have a raft of applications – from the detection of chemical and biological weapons to ferrying casualties safely back to camp using tactical navigation.
They could also form logistics road trains where one manned vehicle could be followed by a convoy of unmanned vehicles. A series of vehicles could also be programmed to automatically carry out repetitive tasks such as digging and obstacle clearance with one supervisor overseeing many operations, thus reducing manpower levels and offering the prospect of greater efficiency and accuracy.
Autonomous vehicles could have a range of uses in the commercial sector, particularly in oil and gas, construction and nuclear decommissioning industries and, indeed, in any potentially hazardous environment.
Under development
The QinetiQ team has already devised a system that undertakes simple programmable operations. Using this technology, on board sensors scan the terrain to produce an obstacle map providing information on the size, orientation, shape and range of obstacles.
Using this data, the vehicle’s navigation system starts to look a ways of negotiating the route in an intelligent and efficient manner, and provides control data to tell the vehicle the speed and path that should be followed. This then allows the vehicle to navigate safely and quickly across unstructured, unmapped, rugged terrain.
The research is still very much in its infancy and there are many technology maturity issues in areas such as communications and sensors that need to be addressed before autonomous vehicles are likely to be seen in production.
“We are already looking at a number of possible sensor solutions, including the use of Ultra Wide Band Radar and Millimetre Wave technology to overcome problems with obstacle detection and foliage penetration”, explains Neil.
Autonomous systems are evolving but they still have a long way to go to be accepted which means that development costs are still high.
For the immediate future, at least, Bill Warren and the team believe that UGVs will continue to rely on visual information of some type being sent from the vehicle to the operator to allow some man-in-the-loop control of operations.
However, the prospect of a fully digitised battlefield with UGVs linking up with unmanned aerial vehicles could be a real possibility in the future.
It is a case of scientists and engineers from within QinetiQ and the wider international research community pulling together to push forward systems integration and technology development.
