So what do we really mean by an autonomous vehicle? A sensible definition would be: “A fully autonomous vehicle is one that does not necessarily require human involvement to make a decision.” But as we can see from all the foregoing, there are varying degrees of autonomy. It is perhaps more useful to focus on a particular application, to examine the state of the art - and then consider where all this is taking us.

One of MIRA’s unclassified projects goes by the snappy title: Mira Autonomous Control Engineering, or “MACE” for short. MACE is high-endurance, fully autonomous, all-terrain platform. It boasts a rugged 4 wheel drive, 4 wheel steer chassis, with a series hybrid powertrain that can carry (and power) a 300Kg payload. This versatile AV is designed primarily to be used for the kind of reconnaissance missions identified by the MOD in the forthcoming “Grand Challenge” – Lord Grayson’s high-profile head to head competition for AV’s with the finals to be held at Copehill Down in 2008. The Grand Challenge aims to produce an autonomous or semi-autonomous technology system designed to "detect, identify, monitor and report a comprehensive range of physical threats in a complex urban environment". The kind of autonomous control and threat detection capability MACE is loaded with to succeed in this sort of task genuinely represents the current state of the art. Aside from the primary speed and directional control systems that each offer fine degrees of fidelity, MACE avoids major obstacles via a laser sensing system. Mission planning capabilities are broken down into 2 main areas, namely patrol and scouting. Whilst ‘scouting’ the terrain classification function allows MACE to relay both topological features and more changeable attributes such as the condition of the ground (hard/easy going) back to mission commanders …vital information that it is not possible to recover reliably from aerial surveillance.

MACE’s intelligent power management system is aware of the energy required to travel from A to B, and it projects the amount of energy required to perform the task. It later compares the actual energy used with its own projections. This information can then be communicated to other similar machines for local tactical benefits. In addition to scouting missions, MACE can be set the objective to patrol an area. This area may be broken into areas of high and low risk, with waypoints overlaid for the vehicle to try and cover, if conditions allow. In high risk locations where quiet operation is desirable, MACE can switch off the IC engine to ensure its position is not easily revealed. It will plan an effective route considering all the permutations including: the state of the ground, any desirable waypoints, the energy required to complete the route and the impact any quiet running has on battery state of charge. The rules of the Grand Challenge require that vehicles investigate an urban environment and look for snipers, mines and IED’s. This drives the requirement for MACE to be able to adapt its route on-the-fly to interrogate ‘interesting’ signals from its senor array.

How is autonomous control achieved? On a practical level, the autonomous control tends to be subdivided into 3 main modules, illustrated in the diagram. The high level module, with all the high-end features such as mission planning, obstacle avoidance and route following capabilities can be a fairly generic unit which could be applied to a range of vehicles, not merely land class vehicles. At the otherend of the system there is a module which understands the vehicles own hardware