Views
3 years ago

Implementing Secondary Swarming - UWE Intelligent Autonomous ...

Implementing Secondary Swarming - UWE Intelligent Autonomous ...

This secondary field

This secondary field could be of considerably less‘broadcasting power’ than the primary target (in factthe power requirements of this secondary field can bereduced further by the use of synchronous shortbroadcast bursts (Melhuish and Holland 1997a,b;Melhuish et al. 1998). An agent unable to detect theprimary target (the signal being less than somethreshold) would attempt to use the same locomotionstrategy but employ the summed secondary fields,generated by those agents which can detect the primarytarget, rather than the primary target itself.It was shown that employing such strategy provided acohesive effect for a moving swarm (Melhuish 1999),where the agents nearer the target occluded theirneighbours further away – even though those ‘near’agents were not comprised of a fixed set of agents.They actually consisted of an ever-changing pool ofagents within the swarm. It was also shown that, for themodel employed, as more robots could find the targetand become secondary field sources, the range of thecombined secondary field increased, which resulted inthe recruitment of more robots. Creating an aggregationbased on this principle was referred to as secondaryswarming.In the context of biologically inspired mechanismssecondary swarming constitutes the employment ofminimal communications combined with positivefeedback. The positive feedback can extend the rangeof the pre-existing environmental template (anheterogeneity in the environment (Holland andMelhuish 1997b, Melhuish et al. 1998), implemented asa primary target, by the addition of a robot generatedtemplate.0.3m 3 giving a net lifting capacity of 93g when filledwith Helium/Air balloon gas, this gas is only 93%Helium making it much cheaper than pure Helium witha negligible difference in its lifting capacity.(A)(B)(C)420mmII. MATERIALS AND METHODSFor this research a group of four autonomous blimpshave been designed and constructed usingmicroelectronics technologies. The blimps have anonboard computer, propulsion system and an infraredlocalisation and communication system. A lithium-ionbattery that gives an operational time of 2½ hoursprovides the power. The following is a brief descriptionof the robots used.The physical implementation of the robot usesa lighter than air vehicle (LTAV) in the form of ahelium filled balloon (blimp). The blimp consists of atwo-panel metalised nylon envelope 96.5cm indiameter uninflated, which once filled has the shape ofa squashed sphere (see Figure 1.A & B) 0.75m indiameter. The volume of the envelope is approximately(D)Figure 1.The Robot Blimp

The blimp gondola is made of lightweightplastic ‘blister packs’ which are commonly foundholding products to cardboard backing. Plastic drinkingstraw arms fixed to the plastic housing attach themotors to the gondola (figure 1.C).Thrusting on the blimp is achieved with threesmall fan units capable of supplying approximately 4gof thrust at full power. Each of these units consists of asmall 2g DC electric motor fitted with a small, 0.3gplastic propeller 5cm in diameter (figure 1.D). Eachmotor is PWM controlled via an H-bridge. One fan wasmounted beneath the gondola for vertical motion.A lightweight ultrasonic ranging system thatgives the blimps the ability to control their height towithin a few centimetres with a maximum range of 3-4metres has been designed. The sensing system has avery good accuracy, however, the dynamics of theblimp and its propulsion system do not allow foraccurate height control. The system uses small separateultrasonic transducers for the receiver and transmitter toenable short-range measurements. The Infraredlocalisation system is based on a design by Kelly andKeating [1996] that gives the relative position anddistance of the other blimps and allows for lowbandwidth inter-blimp communications. There are threetypes of analogue board that make up the infraredlocalisation system: the receiver, transmitter and twosensor boards. These boards are fed from a voltagesupply that has been filtered to minimise the noiseinduced by the high speed switching of the digitaldevices.The design employs the Li-Ion ‘SonyMinidisc’ battery (LIP-12B), which gives 1500mAh at3.6v for a weight of 50g. minus its plastic casing. Thiswould allow an operational time of up to 2.5 hours.The electronics are split into two distinct parts;digital and analogue in order to minimise the couplingof noise from the digital board to the much moresensitive analogue boards. The digital circuit consists ofthe power generation, motor drives, frequencygenerators and a microcontroller all of which work athigh frequencies producing a lot of noise and therefore,electrical interference. All the circuits are manufacturedfrom 0.5mm fibreglass board, which is one third thethickness of a standard circuit board. This thin boardcoupled with the very compact layout and the use ofsurface-mount technology helps to reduce the weightby a considerable amount.The motor drivers are HIP4020 full bridgedrivers and are capable of delivering 500mA in anSO20 surface-mount package.The frequency generators(figure 3.8a) used by the transmitter and receiver usedirect digital synthesis (DDS) technology to producenear perfect sine waves with minimal externalcomponents. The ICs used are AD9832s that arecontrolled via a three wire serial interface and come invery small (TSSOP) surface mount packages.The heart of the system is an 8-bit ArizonaMicrochip PIC 16F877 microcontroller, which runs at afrequency of 20MHz giving ample computationalpower. This microcontroller has 8Kbytes of onboardflash memory, which can be programmed in-circuit viaa three-wire serial interface, three 8-bit digital I/O portsas well as an 8 channel 10-Bit AD converter. The unitalso features two PWM and three timer modules.III. EXPERIMENTAL DETAILSThis set of experiments attempts to discoverwhether a collective strategy can increase theperformance of a group of robots when homing on astatic beacon. Two different strategies wereimplemented and compared; pseudoswarming (whenrobots act as completely independent agents) andsecondary swarming. When pseudoswarming, therobots individually home in on the beacon with nointeraction between each other. However, they can stillgive the impression of collective behaviour as theymove toward the beacon. With a small change to thealgorithm the robots can be made to interact with eachother changing the way in which they home on thebeacon.The experiments were conducted in the experimentalarena with the beacon fixed in one corner at a set heightof 1.5m and the robots started in the opposite corner(see figure 2A below) to give a separation distance ofapproximately 10m, (just within the blimps maximumsensing range.) Four robots were started together andtheir positions recorded at one second intervals by theoverhead camera system. A time limit of two minuteswas placed upon each trial to limit the amount of datacollected, it was expected that on a good run a robotwould reach the beacon within 30 seconds if it couldinitially detect the beacon and did not lose its signal.The following secondary swarming algorithm wasimplemented on each robot:

Distributed communication in large swarms of autonomous ... - EPFL
From Swarm Intelligence to Swarm Engineering - EPFL
Implementation of the European autonomous framework ... - ETUC
Implementation of the European autonomous framework ... - ETUC
Modelling Swarm Robotic Systems - UWE Intelligent Autonomous ...
I-Swarm: Intelligent Small World Autonomous Robots
A Swarm Intelligence inspired Autonomic Routing ... - IEEE Xplore
collective perception in a swarm of autonomous robots - laral
Challenges. PDF - UWE Intelligent Autonomous Systems Lab ...
Micromanipulation, communication and swarm intelligence issues in ...
A Physicomimetics Control Framework for Swarms of Autonomous ...
Analysis of Algorithms to implement Swarm Bots for Effective Swarm ...
Brownian Motion as Exploration Strategy for Autonomous Swarm ...
Swarm Robots for Autonomous Thrash and Garbage Removal
Swarm Intelligence – W12: Segregation, Aggregation, and ... - EPFL
Formation Flight of Swarms of Autonomous Vehicles In Obstructed ...
Implementation of Control Systems for Autonomous Robots
Artificial Swarm Intelligence vs Human Experts
Design and Implementation of an Autonomous Aerial Vehicle for ...
Artificial intelligence: Riders on a swarm | The Economist
Swarm Intelligence Market
Teaching and implementing autonomous robotic lab walkthroughs ...
Implementation of a Generic Autonomic Framework
Towards the Application of Swarm Intelligence in Safety Critical ...