Shadow Deminer title

Document created: 16th August 1999
Last Modified: Wednesday, 21-Aug-2013 17:10:20 BST
Icon saying Projects & Research

Click here for Shadow home pageClick here for Liberator pageClick here for Zephyrus pageClick here for Mine clearance pageClick here for Other projects page

This research project is no longer active.

Contents to this page


The concept is one of a vehicle capable of traversing an anti-personnel minefield carrying mine detecting sensors or video cameras. The vehicle must be able to traverse rugged terrain and degrade gracefully in the event of damage.

The proposal is for an eight legged vehicle with emergent walking behaviour using pneumatic actuators and local materials where possible. These factors contribute to the simplicity of the basic vehicle and low cost if destroyed. The vehicle is designed to protect the mine detecting sensors at the expense of the vehicle if necessary. As a result, only this detection equipment is placed in protected spaces, and the legs in particular are sacrificial.

3D model of Deminer - 'Infini-D 4.5'


A fully equipped mine-detector vehicle needs to be able to complete a variety of possible missions...

  1. Area detection: 100% detection is needed. The mines are marked for later manual clearance, so a high false positive rate is acceptable, even preferable. A search rate of 90 mins. per square metre is acceptable.
  2. Reconnaissance: A wide area is loosely searched using a video camera. The purpose is to confirm the existence of mines and determine their type.
    A speed of 1 kmh. minimum is needed, preferably more.
  3. Vegetation clearance: Up to 50% of a deminers time is spent in vegetation removal so this is no small part of the operation. Trees and shrubs under 20 mm. diameter are cut. Burning is not acceptable as the burnt plastic and unexploded TNT can pollute the soil. This area is not currently part of the project.

Thus the vehicle part of the project has the following requirements...

  • To be able to traverse very rugged terrain at 1 kilometre/hour for positioning under radio control and long range searching. At the very least a 30 degree slope should be traversable for early prototypes.
  • To execute a search pattern that can cover 1 square metre in 90 mins. without leaving a gap larger than 3 cm. This should all be possible in rugged terrain.
  • To have a ground weight less than 5 kg. per leg/wheel.
  • In the event of disaster, a small mine strike should only impede the vehicle, allowing the vehicle to return home. For larger mines and jumping mines the vehicle can be disabled, but the expensive mine detection sensors should survive intact.
  • To be able to accurately dispense markers that are undisturbed by the vehicle's later manoeuvres.
  • To have a loiter time of 1 hour in the field with all sensors operating and a range of 2 km. when traversing terrain.
  • To undertake a local search pattern without human control. The only intervention should be if the vehicle gets seriously stuck or has to signal an alarm because a section of terrain cannot be reached.


For reconnaissance missions the navigation and other operating procedures are dependent on the terrain and available sensors and so will not be discussed further here.

For clearance missions the search pattern has a large impact upon the vehicle design.


The actuators are adapted 30 mm. Shadow Air Muscles which are both light and cheap and simple enough to be low technology manufactured (there is no machining, but they need labour). Four muscles control each leg arranged in opposing pairs in the manner used in the Zephyrus robot.

It is probable that a large number of muscles will suffer shredding in the event of an explosion. The muscles are designed to be easy to attach and replace, however.


This is one of the main research areas associated with the vehicle. The vehicle has to produce a walking gait in rough terrain. In addition it has to develop a gait with one or more legs missing.

The initial prototypes and computer simulations will have eight legs. The highest speed will see one leg on each side in the air at any one time. With 5 kg load allowed for each leg, this gives a maximum weight of 30 kg for the whole vehicle.

The length of the legs imposes restraints on the type of terrain. Some manoeuvring will be required for trees and hollows. It is not necessary for the vehicle to move very quickly. Although the legs are long, the actual steps will be short to prevent leg interference. The requirement of at least 1 kilometre per hour for reconnaissance imposes a movement rate of 1.1m per second per leg-tip at full speed. In search mode covering a square metre in 90 mins. (Approx.) gives a half hour for 1 metre of forward movement. The steps would be much shorter in this case; this also permits the leg to be rested on an obstacle, and shorter steps taken. Typically ten steps could be taken to cover one metre with a speed of 1-4 mm. per sec. of forward movement for those legs on the ground.

Control Electronics

Following the classical subsumption architecture for the design of autonomous vehicles, we divide the control electronics up into layers in this way:

  • At the lowest level, the muscle pairs in each leg use the strain gauges on the muscles to maintain posture.
  • The next level up is the interconnection of neighbouring and opposite side leg pairs to generate the walking gait.
  • The control computer makes use of odometric information extracted from leg sensors and the ground-pointing camera navigation system and local map building to decide the direction of the walking motion.
  • The long range navigation or radio control system provides strategic direction of the vehicle, including the determination of when terrain is unreachable or impassable.

Every opportunity will be taken to integrate these components with the specific mine-detecting systems available. Thus these systems must be flexible and able to receive outside stimuli.

Other constraints on the electronics include survivability against severe shock, simple maintenance and resistance to high temperatures and humidities. The vehicle must also protect sensing equipment as far as possible from fragments and rain.

Energy Sources

The main energy source will be either a petrol or diesel engine. This will eventually supply a combined compressor and generator topping up batteries. The initial prototypes will be built from standard components.

Fuel Tanks

There are two options for the fuel tanks. Either to mount them in the protected spaces or to mount them sacrificially. The sacrificial arrangement (to be experimented with) is to mount two tanks on poles, forward and aft, a metre clear of the vehicle and the sensor probes. Fuel is gravity fed down fire retardant hoses wrapped around the poles. The tanks can be cheap plastic containers replaceable locally. It is probable that both tanks will be ruptured in the event of an explosion. Hopefully this will result in the fuel tank being torn from its mounting and thrown clear, taking the flammable fuel clear of the vehicle sensors. There should be enough fuel in the fuel lines for the vehicle to move clear of any burning wreckage assuming that the robot is still mobile.


The compressor and generator need to be of simple design fuelled by petrol or diesel. Ease of field maintenance and repair is vital. The compressor does not need to supply a large air reservoir, nor does it need to provide a high pressure; therefore it can be reasonably light. If possible components will be duplicated for survivability. All reservoirs and casings will be plastic, where possible, to reduce weight and to prevent metallic fragments contaminating the minefield.


Electricity has to be supplied to the mine detection sensors, navigation electronics, radio gear and air valves. Truck batteries are too heavy, so 6 V. motorcycle batteries (lead-acid) will have to be used. It is unlikely that anything else will be available locally.


The three main defences are:

  1. Low ground weight: The vehicle will exert a maximum weight of 5 kg per leg.
  2. Leg weakness: The legs are up to two metres long with a deliberate weakness near the body. In the event of a mine being detonated, the leg will be sacrificed and the vehicle will attempt to move a short distance to a cleared area. The leg can be easily replaced from locally-available materials.
  3. Layout: The heavier and cheaper equipment is placed to the sides. This is to protect the expensive sensors from fragments dispersed horizontally by jumping and trip-wire mines.


There are two navigation problems: local whilst searching, and long range. The following are statements of design goals. Intermediate prototypes will have simpler systems.

Local Navigation

Local navigation will be by ground pointing camera or sonar. Sonar is preferable as this also tells the vehicle how high off of the ground the sensors are, but a video device is more robust. The ground pointing sensor thus gives the same information as odometry and should be adequate for the last ten to fifteen metres of travel.

Long Range Navigation

Long range navigation could be by differential GPS, which is accurate to 1 to 3 metres resolution. Otherwise the operator has to walk 100 metres behind the vehicle using radio control.

Mine Sensors

There are several options as the vehicle is designed to be a general platform for a wide range of sensors many of which are not yet available. The vehicle body can be positioned at any height from 10 cm. to 60 cm. or more and makes available a 40 cm. wide, 1 metre long platform for sensor placement.

Possibilities include: Metal detector,The Alberta probe and Ground penetrating radar.


Real time control of walking
Marc D.Donner, ISBN 0-8176-3332-4
Sensor technologies for the detection of anti-personnel mines, A survey of current research and system developments.
ISMCR'96, Bruxelles. May96, pp.6-10, Bertrand Gros and Claudio Bruschini, EPFL-LAMI, DeTeC, Laussanne, Switzerland. (find this Internet link)
A mechanical means of mine detection, Final report. Univ. of Alberta, dept. of Mech. Eng.
David Skaley, Jeff Hemsing, Ken Fyfe. Nov96. (link to Alberta university)


Many thanks to those companies and organisations that have helped the project get off the ground. Many thanks to Cambridge Adaptive for the supply of controllers.
Soroptimists International for start-up funding.
Special thanks also to the Mine Advisory Group, the Halo Trust and DERA who have provided crucial consultation.

© Marcus Baker and The Shadow Robot Company 1999

An image saying Millenium Products   Click here to goto Shadow Robot Company main website

©Copyright Shadow Project.

  An image saying Engineering Industries Association