Self-Organization in Large Populations of Mobile Robots:

5. Technical Assessment

This chapter is a brief discussion of sensors and communication methods applicable to the Army-ant scenario. In Army-ant robots, there would be five main modules for interaction with the environment:

  1. communication module
  2. beacon detection module
  3. obstacle avoidance module
  4. locomotion module
  5. force/position detection module

We do not discuss force sensors, inclinometers, or locomotion devices --which are necessary for modules iv and v-- here. Present technology used in robot manipulators and mobile robots is adequate for application to Army-ant robots. We will discuss several types of sensors that can be applied in realizing the first three modules. Our proposal of the devices and methods specified here is based on the scenario described in Chapters 3 and 4. The list of the devices to be used in Army-ant robots will take shape in the process of realizing these robots. Of course, the main decision factor will be the cost per agent.

5.1 A Communication Method: Binary Coded Signals

As explained in previous chapters, Army-ant robots need to communicate with each other in order to share information available to some of the team members. Binary coded signals can be used to inform teammates about the goal and problems during the pallet transport. Request signals described in Chapter 4 (warning, wait-for-me and status of agents) can be coded and transmitted using short range radio communications. It is possible to address a specific agent using this coded signals. For example, an agent can broadcast wait-for-me signal to all robots. The structure of the coded message can be arranged so that only the robots whose behavioral condition allow them to respond, can stop for their (would-be) teammate. Those agents may continue when the same signal is not received again in some predefined time interval.

Besides these messages, it is also possible for agents to send information that is necessary for the coupled VDP oscillators method described in Chapter 4. Ring formation can be achieved by exchanging data identifying the agents. However, oscillator (or other continuous analog) signals will probably be transmitted by RF links. To keep the number of necessary channels to minimum, agent beacons can assume different roles at different phases.

Agents may share a single channel provided that they can detect whether the channel is busy or not. Genovese et al, defined the necessary strategy for such a communication method in [16]: Each robot would have to determine the presence of the carrier signal using a radio transceiver module before transmitting any data in order to avoid "collisions," and a control bit added to the end of the binary coded signal have to be used to detect errors to achieve "robust" communications.

Digital communications via RF and IR links are becoming cheaper and there are a variety of companies providing solutions. Several radio modems that use fixed or spread spectrum techniques are available. Some of those devices can even be directly plugged into the serial ports of the microprocessors. Necessary devices for an RF link (i.e., FSK Data Modulator/Demodulators, Digital fixed Frequency Transmitter/Receivers, Computer Controlled Synthesizers, etc.) can be obtained around $50-100 each.

The amount of the data transferred by agents at each transmission would be quite small, considering that we can fit all data needed into a few bytes. Therefore, the time interval between transmission attempt and end of transmission should be short.

5.2 Sensors

In addition to the communication system for complex data transfer, Army-ant robots have to be equipped with sensors to detect obstacles and beacons for goal search and navigation. Acoustic and radiowave (microwave) sensors are the two most commonly used sensors when there is a need for computing the distance to an obstacle. These sensors measure the time between emission and reception of the wave (time-of-flight method) to calculate the distance. They are both "active" sensors, and can be used as short-range proximity sensors for obstacle avoidance and teammate detection.

Although the time-of-flight method is useful to detect obstacles and teammates in the Army-ant scenario, agents also have to detect the distance to a beacon --where time-of-flight methods are impracticable. However, there exist some short-range radars (radiowave sensors) realized by RCA, and a system using reflected energy instead of the time of flight is presented in [1]. This method should enable Army-ant robots to calculate their distance to the pallet and teammates.

Acoustic sensors have several advantages over radiowave sensors: they are smaller and not sensitive to the nature of the target, and they can be used for underwater applications. On the other hand, for space applications, radiowave sensors are the only choice, their physical principle being usable in space and for large distances.

Ultrasonic sensors are likely to be used in Army-ant robots. Ultrasonic sensors may use a large frequency range (100 Khz-1Mhz) very effectively. Resolution on distance measurements can be as low as 0.01 mm in air [23]. Qualities of ultrasonic sensors important to robotics applications are sensitivity, dynamic range, linearity, space and time resolution and signal processing. However, they are not flawless; acoustical waves projected onto a surface under an angle are deflected like a light by a mirror, and no echo returns to the receiver. They are also susceptible to wind --which may be an important factor in outdoor applications.

The most widely used acoustic sensor is the Poloroid acoustic sensor. It has a range between 26 cm. and 10.7 m, and +-%1 resolution over the entire range, and requires 6.0 V DC. Its acceptance angle is 20 degrees. Transducer and ranging modules weight 8.2 and 18.4 gm respectively. There are also other sonar ranging modules under $50. With these specifications, sonar sensors seem to be a feasible choice.

One of the most important problems in Army-ant robot navigation is to detect the direction of the beacon signals. Again, two types of signals can be used for direction finding: radio or ultrasonic waves.

The direction of ultrasonic beacons can be found by phase comparison between pairs of acoustical receivers. A navigation system developed in MEL, Tsukaba, Japan uses three independently rotating pairs of receivers to find the direction of three ultrasonic beacons [2]. However, the application of this idea to Army-ant robots seems difficult since they are to carry payloads. Other than rotating acoustic sensors, geometrical arrangements of multiple sensors or an array of acoustical transducers are well-known methods for environment perception. These methods can also be adapted to Army-ant robots for beacon detection purposes. Arrays of acoustical transducers require interpretation of the signal received, but this can be realized by a small microprocessor. In [13], it is stated that acoustical phased arrays are low cost, and are useful for applications with mobile robots to give a rough perspective of the environment. Therefore, such a system must be adequate to detect the direction of a beacon.

Another type of sensor that we do not consider is laser range finders. Although they are more precise than other range finding devices, the inexpensive ones have maximum accuracy of 1/4 inch, and cost approximately $600, four to five times more expensive than sonar range finders. A more futuristic approach is to use gas sensors. Pelosi and Persaud state that gas sensors, based on polymers of aromatic and heteroaromatic molecules that exhibit electrical conducting properties, can be used to detect specific odors [29]. Presently, reducing and oxidizing gases can be monitored by the impedance of SnO2, ZnO, TiO2, and other oxide surfaces. A gas chromatographic air analyzer was fabricated on a silicon wafer using solid state electronic techniques [36].

Considering the number of robots, there will be a large number of signals on all the used channels. While it is possible to 'queue' the data transmission in Digital RF or IR communication, beacon detection may be highly difficult due to the number of signals emitted and/or reflected. It is difficult to discuss the performance of such a scenario before realizing Army-ant robots. On the other hand, we know that a certain amount of noise, whose emergence can be regulated by the communication system, can even be advantageous for pallet search as in the foraging process of ants.


Back to Thesis Main Page

To Chapter 4

To Chapter 6