SlugBot: Towards True Autonomy



Pictures of the SlugBot (as of May 2001)


Most mobile robots are not truly autonomous; most operate in simplified environments. Almost all non-industrial robots still require a helping hand from humans, e.g. battery charging, the odd push if they get stuck, obstacles that are suited to their sensors, etc. The notable exceptions are smart missiles, satellites and torpedoes which carry enough fuel and computational resources to successfully complete their missions.

On the other hand, even the simplest animals are self-sufficient, both in terms of information processing and energy. The aim of this project is to build a robot with animal-like self-sufficiency in both information and energy. We don't expect to be able to match the speed and performance of a cheetah chasing a zebra, within the time frame of this project, so we decided to chase something slightly slower ... slugs. Apart from their relative ease of capture (compared to zebras), slugs were chosen because they are a major pest, are reasonably plentiful, have no hard shell or skeleton, and are reasonably large. It is also more technologically interesting to catch mobile prey rather than just grazing on plants.

Of course, the organic resources, or food (i.e. slugs), will have to be converted to a form of energy that is useful to a robotic system. We propose to convert the organic material to electricity using microbial fuel cells. Agricultural fields of winter wheat offer a suitable test bed for the robots because slugs (Deroceras reticulatum, Arion ater ater and Arion ater rufus) are plentiful, with up to 200 per square metre. Slugs are mainly active at night, especially just after sunset and just before sunrise, so the robots will have to be active at these times, and resting in order to conserve energy during the day when most slugs are underground. Since energy conservation will be of prime importance, and moving heavy items over soft ground will consume large amounts of energy, the fermentation vessel, engine and generator will be stationary. One or more robots will deliver slugs to the stationary charging system and obtain power from it. In order to minimise the amount of movement each of the robots will be equipped with a 1.5m long arm, mounted on a turntable, which will be used both for detecting and collecting slugs which is now working under closed loop computer control, see the following mpeg movies: the arm moving (1MB) and the arm scanning (1MB) and a longer but lower resolution clip of the arm scanning (1MB) If you think that looked easy, take a look at us trying to manually control it (3MB) and the outtakes (2.5MB) of our attempts.

Although slugs are relatively difficult to see under daylight with a naked eye, they can be relatively easily detected with a CMOS image sensor under red illumination. Under red light grass and other vegetation appears dark whilst slugs appear quite brightly. The three images below slow the same slug, which is a 32mm long Deroceras reticulatum, under different lighting conditions. The left image is illuminated with white light, the middle image is illuminated with red light, and the right image is the middle image but with a simple threshold applied. On the left of the left image is an area of high slug slime, this is also filtered out by illumination with red light. Unfortunately stones also appear quite brightly as well.

The final stage is the elimination of stones by the identification of bright patches that are of the correct size and shape for slugs. This is achieved by finding the biggest connected bright patch in each image and by calculating the ratio of bright (slug/stone) pixels to dark (background) pixels within a boxed area around the bright patch. The length and width of this box are equal to the length of the patch. The rationale behind this is that most of the time slugs are stretched out and are long and narrow and hence fill a small proportion of the box, whilst stones tend to be much rounder and fill a much higher proportion of the box. This system has the advantage of being invariant to the orientation of the object under examination and within pre-set bounds of its size.


Images of a slug under different lighting conditions


Once a robot has collected a load of slugs it will need to return to the static fermentation station before its battery is fully discharged. Locating the station will be achieved by using a combination of the Differential Global Positioning Satellite (DGPS) system, conveniently set up for us in advance by the American military, and an active infrared localisation system. DGPS, which offers sub-meter resolution, can also be used by the robots for mapping out food resources. Obstacle detection will be achieved by a combination of ultrasonic sonar and bump sensors.

Current Status of Project

Currently the prototype can move, scan, detect and almost collect slugs whilst ignoring stones; in the laboratory under conditions similar to those found in real agricultural fields, see the following mpeg movies: short clip (3.3MB) and longer clip (10.5MB) (please note that no animals were harmed during these experiments - the slug is plastic) Unfortunately the current gripper cannot collect slugs/sausages that are in deep dips in the ground. Work has now started on the development of a usable microbial fuel cell.

For further information about the SlugBot please see my publications.

Back to Ian Kelly's home page