Você está na página 1de 9

Autonomous Swarming Robots

Design Proposal

Group Leader: Brian Simms Evan Ng Dien Nguyen Joseph Salomon Brian Woo Senior Advisor: Yan Meng Professor: Bruce McNair

I pledge my honor that I have abided by the Stevens Honor Code.

I. Abstract Robotic swarms have capabilities to do things single robot systems cannot. The inherent parallelism and redundancy of the system also give large groups a distinct advantage. Furthermore, smaller robots are less expensive and can navigate areas large robots would have difficulty with. The proposed project is to design and implement the hardware platform and develop robot control and map-building swarming algorithms, for a modular miniature autonomous mobile robot swarm. The hardware will be modular so that damaged components can be replaced without replacing the whole system. This will also allow sensor packages to be changed before runtime with little effort, making the swarms collective perception more dynamic. In short, the group will be making a complete modular autonomous swarming mobile robot system from conception. The approach will be to design each robot to be low cost, low power, and use readily available components. Multiple robots will be produced, and they will work together in a swarm to perform various tasks such as map building and object searching. II-1. Introduction There are several excellent examples of miniature robots that have been previously constructed. The autonomous mobile robot ALICE, developed by the Swiss Federal Institute of Technology Lausanne [1] is one of the more simple examples. It uses a PIC16 series microcontroller infrared sensor, and a bidirectional radio communications module to perform many tasks. It can localize itself and map build to a certain extent; though inter-robot communication has been basic, it is an excellent example of a low power, low cost robot. Kenyon et al. introduce a small, cheap, and portable reconnaissance robot with a unique spherical chassis in [2]. This robot allows for modular sensor hardware so the robots perception capabilities can be changed before runtime. The robot uses a PIC18LF458 8-bit microcontroller for processing and 2.4GHz wireless for communications. This will be a good reference for a robot with a modular sensor interface, which is part of our design plan. The iRobot corporations SwarmBot is equipped with both an ARM processor and a 200k gate FGPA. The infrared sensors, light sensors, and camera help the robot navigate during such things as dispersion techniques, as discussed in [3]. This is one of the best examples of swarming robots, and will likely be a valuable resource. There are many things to consider when designing both the hardware and software for swarming robot systems. The design goals of this project are as follows: Provide a modular hardware and software architecture. Individual modules should be easily modifiable and replaceable. Provide a high performance microcontroller in a common and well supported architecture. Open source compilers and debugging tools should be available. Provide a reconfigurable hardware component. Again, this component should be well supported with software tools and documentation. Provide wireless communication. An inexpensive, off the shelf module is ideal.

Provide a large number of interfaces for sensors. Sensors such as infrared proximity sensors, cameras, sonar sensors, and bumper switches need to be connected while leaving room for future expansion. Low power. Mobile robots generally run off of small batteries. The system should not consume more than 1000mW. Low cost. Using high end microprocessors or FPGAs is out of the question. PCB design should fit the design rules of low cost, fast turn prototypes. Total cost should be under $400USD each in single quantity. Small Size. The complete system should be approximately 3 inches cubed. Low risk design. Proven designs and off the shelf hardware will be used whenever available. Ease of fabrication. A reasonably skilled user should be able to assemble the necessary components into a working system.

The need for a modular interface has already been examined. It will be needed for quick adaptation of the system. The importance of a well supported processor cannot be understated. Many difficulties in development are due to simple lack of documentation. Using a popular component will help alleviate this; further, since the cost of the project is limited, open source tools are a must. For any autonomous robot, machine learning capabilities are a highly desired feature. A reconfigurable hardware component will be part of the design goals in order to satisfy this. Any swarming robot needs many sensors to understand the environment; this includes wireless for inter-robot communication. Finally, autonomous nature of a robot is limited by the amount of time it can stay in the field. This is why the robot will be low power. The following sections will discuss in detail these requirements, as well as our approaches to them. Less technical objectives will then be discussed, such as out financial budget and schedule. II-2. Design Requirements The design requirements state the engineering and hardware baselines that will ensure a successful project. The group intends to build multiple identical swarm robots for our final project. With this in mind, cost and use of identical parts must be taken into account. The robots will be relatively small and lightweight, allowing ease of movement and maneuverability. The robots must have a method for code testing and debugging in the throughout all the stages of the project. In order for the robot to interact intelligently with its environment, each will be equipped with multiple sensors in order to assist in the localization, hazard detection, and target discovery aspects of the design. Area and distance sensor alternatives must be weighed. Furthermore, the robots must be designed in such a way that will prevent them from running into boundaries or getting stuck in corners. Each unit will be fully functional and independent from the swarm, but at the same time able to coordinate and work with the other robots. This will require the implementation of various robotics and engineering technologies. To handle the communication aspect, each unit must be

able to have wireless networking capabilities, allowing it to send and receive information to a host location as well as to the rest of the swarm. From a numerical standpoint, there are no any feasible parameter values or design objectives at this point. However, the group will aim to design the robots to be low-cost, low powerconsumption, and modular. In general terms, the group would be planning to have a swarm of robots perform tasks such as traversing a room, creating a map, or finding a set of objects faster and more efficiently than a single robot, but no numerical data has been collected at this time. The group is not aiming to upgrade a given set of parameters, but create an overall efficient system of coordinated robots.

II-3. Design Approaches Based on the previous requirements mentioned, alternatives to meet our goal were considered. The main design requirements consist of mobility, localization, sensing, and communications methods. Mobility There are numerous ways to implement and achieve our objectives. Initially, the group had the option of selecting ground or air-type robots but ultimately chose ground-type robots due to the time allotted for construction and programming development, resources available to the group, and cost allocated for design projects. One alternative for mobility is to have walking-type robots. The group would have leg-type mobility. This would allow for the simulation of human-behavior mobility to scout the surrounding area. The robot will move relatively fast due to its nature. The mobility of the robot will be done with a total of 4 legs with each leg processing a motor to move its mechanical features and programming aspect to regulate and move the mechanical parts in an orderly fashion. Some difficulties that may be encountered would be the programming the leg joints to work in conjunction with the motors because the joints would be have a 360 motion, the legs and motors must be programmed precisely in order to move the robot in the proper manner. A second alternative for mobility would be to have tank-type robots. Since tank treads are able to move on a multitude of terrains, this type of robot will eliminate the need to go around small obstacles. Since the robot will use treads as its mobility, programming aspect of mobility will be simpler than having leg-type robots. Localization In order for the robot to be truly autonomous, the robot has to know where it is via robot localization. The robot has to know where it is in order to execute its task. In order to localize, the robot has to obtain some reference point and which allows the robot to determine its relatively location from which the robot can gauge its environment and surrounding. That being

said, the robot has to know its location. The group evaluated the global position system (GPS) as one way to determine its absolute location. The GPS is a technology used for outdoor navigation developed by the Department of Defense. The GPS system consists of 24 satellites in geo-orbit position that transmits RF signals. The robot can use the trilateration method that allows for calculation of location with an accuracy error. The accuracy error, the group deemed, isnt as important since the robot communicate with each other and that allows the robot to localize in respect with each other. When they localize with each other, the searching algorithm allows the robot to cover more area with much more efficiency. Another type of localization which is most commonly used type is called odometry. This technique is accurate in the short-term and inexpensive. This technique uses wheel revolutions data which are the converted into linear displacement relative to the floor. The drawback to this technique is that it is volatile to error; meaning that, if there is a slight error in calculation, then the entire navigate is skewed. Another type of localization is the use of a compass. Since there were accumulated errors when using odometry, the use of a compass would resolve this issue. According to an article by Borenstein et. al a drawback to this method is when the operating robot goes near any power plants or steel objects, the results are distorted. Map-based positioning is a third type of localization that the group found. This technique can be categorized into two parts: a pre-saved map, or creation of a map. In the first case, the robot would create a scan of the current environment and cross-list that environment with its own database. If there is a match, the robot can calculate the exact position in the surroundings. In the other category of map-creation, the robot would create a map of the environment using its available sensors and create a map using exploration strategy maximizing the amount of explored area and minimizing the time to generate it. The use of sensor fusion would assist in building the new environment. Sensors Not only is the mobility important in a robot, how the robot perceives the surrounds is a necessity as well, namely the sensors. There are various sensors that this robot can use in order to detect the objects in its surroundings. One type of sensor is sonar. This sensor sends out sound waves which are then detected after they are bounced off objects. Afterwards, the time delay between sending and receiving the sound wave is collected and further processed to find out the object. Some drawbacks of sonar is that the sound wave can be reflected back creating the same angle (specular reflection); thus causing errors in data. Another type of sensor is infrared. These commonly used infrared sensors are similar to sonar in that an emitter sends out a signal and the receiver collects the data. The generated time difference would be obtained and the distance between the robot and object would be calculated. A third type of sensor would be laser. Lasers have the greatest range and precision than sonar and infrared. But the drawbacks to lasers are that they are heavy and utilize a great amount of power. Another issue is that lasers are also expensive as well.

These are the various types of sensors that can be used in order design. Not only one sensor can be used but multiple sensors working in conjunction called sensor fusion in order to provide the most efficient and effective result when investigating an environment. Wireless Communication A wireless system replaces the need for a physical serial connection by establishing a layered connection between robots. One of the most crucial parts of robot is the need for a robust and reliable communication link between the robots. The communication is asynchronous so it will take huge power during transmission and monitoring. There are several solutions available to enable communication between the robots which the group will compare the advantages and disadvantages of each solutions. The available options are radio frequency, Bluetooth, and 802.11. Although they serve as the same purpose, they have advantages and purposes. The first solution is radio frequency since it allows information to transfer at high speed and over long distance. The huge advantage of RF is that it is easier to implement and readily available for a cheap price, the low power consumption. In the technical sense, raw data is transmitted via RF for the microprocessor on the robot to interpret and execute. As far as interference, RF operates in 2.4 GHz and has multiple channels to minimize interference from outside sources that runs the same frequency. However, the disadvantage is that there are limited channels and it may be prone to inference if there are too many swarm robots. The second solution is Bluetooth. Bluetooth is a radio frequency standard that was designed primarily for short range communication between devices such as mobile phones, laptops, PCs, printers, digital cameras and video game consoles. Bluetooth, known as 802.11, uses the 2.45 GHz frequency band, which is a widely used frequency known as ISM (Industrial, Scientific and Medical). to avoid interfering with other protocols that use the same band, like wi-fi, the Bluetooth protocol divides the band into 79 channels (each 1 MHz wide) and changes channels up to 1600 times per second. This technology allows two or more Bluetooth devices to be used at the same frequency at the same time. The maximum allowable devices are up to 8 at the same time within the range of 10 meters. It is a primary designed for low power consumption and low price. For the purpose of this swarm robot, Bluetooth may be not be suitable because of the short range, however it does satisfies the power requirement. The last solution that the group looked at is the 802.11 protocol. Another name for 802.11b is called wireless fidelity, or Wi-Fi. It is an IEEE standard for Local Area Networks without a physical connection that operates in the 2.4 GHz band. The IEEE expanded on the original 802.11 standard in July 1997 to create the 802.11b which the speed is comparable to the traditional Ethernet at 11 megabits per second. And operating range is 30 meters. The Wi-Fi is the most popular standard that dominates the wireless home networking due to the high speed it offers at an affordable price. The Wi-Fi is mostly used as an extension of LAN. Today, most home office and university uses a derivation of this standard to provide wireless LAN. It is where a wireless network is formed through an access point and clients connect to it. The advantage of the 802.11 is that it provides very long range but it is very high power consumption. Since it operates at an FCC unregulated frequency, the standards are susceptible

to interferences from each other and with those who share the same band such as microwave oven, cordless phone, etc. Below is the table that summarizes the two wireless standards. Table 1: Wireless Standards Technology Range 802.11B ~30meters Bluetooth ~10meters Radio Frequency antenna dependent

II-4. Financial Budget The proposed costs for this project are presented below. With this projects context, many overhead, manufacturing, and testing costs are not encountered. Table 2 shows the possible parts and costs for a single autonomous robot. It utilizes standard products and pricing taken from circuit and electronics distributors. Leeway was included to allow for the option to choose different component alternatives for the sensors, communication, etc. The prices reflect low-volume purchases. Higher volumes would show significant discount.

Table 2: Preliminary Parts List per Robot Part Price Microprocessor $15-30 Movement method $100-180 Wireless Communication $20-40 Memory $30 Sensors $40-80 Imaging $60 Price per unit $265-420

Many cost factors will not be associated with this project, as many will not apply. Labor will be provided by the group and will utilize existing lab equipment. Testing equipment, such as voltmeters, ammeters, soldering irons, and a hot air station are readily available for use. Computers are also ready for use. Because of the close proximity of the equipment, travel and destination costs will not apply. There is no customer discussion at the moment as well. Rent and utilities have already been paid for, regardless of the project, and are provided by the school. Therefore, the main costs for this project are the actual robot components and parts.

II-5. Project Schedule Figure 1: Project Schedule

The above figure shows the proposed schedule for the project. The different design approaches will follow this timeline, as only the design alternatives or components will change. The proposed schedule durations will still remain. At this point in time, the tasks are not broken into much detail. As the project progresses, tasks will be refined and durations adjusted. As can be seen, hardware and software development are the two main tasks, with underlying tasks clarifying them. Brian Simms and Joseph Salomon will work mainly with the hardware design, development, and implementation. Evan Ng, Dien Nguyen, and Brian Woo will delve into the algorithm coding and testing, with Salomon supporting as needed. All members will produce deliverables.

III. Conclusion The objective of this project is to design, develop, and implement an autonomous robot swarm in order to perform a collective map-building task. The motivations for this endeavor are the cost, parallel-processing, and system redundancy benefits that are inherent within collective cooperative units. A single task, such as the exploration and mapping of an unknown environment, can be completed quicker and more reliably by a coordinated group instead of a single, more expensive robot. These swarm robots will be designed to be modular, small formfactor, low-cost, low-power consumption, and easily assembled.

IV. References
[1] G. Caprari, K. O. Arras, R. Siegwart. The Autonomous Miniature Robot Alice: from Prototypes to Applications. Swiss Federal Institute of Technology Lausanne (EPFL). [2] S. Kenyon, D. Creary, D. Thi, J. Maynard. A small, cheap, and portable reconnaissance robot. Sensors and Command, Control, Communications, and Intelligence (C3I) Technologies for Homeland Security and Homeland Defense IV, Proc. of SPIE, Vol. 5778, 2005. [3] J. McLurkin, J. Smith1. Distributed Algorithms for Dispersion in Indoor Environments using a Swarm of Autonomous Mobile Robots, iRobot Corporation. [4] J. Borenstein, H. R. Everett, L. Feng, D. Wehe. Mobile Robot Positioning Sensors and Techniques. Journal of Robotic Systems, Vol. 14 No. 4.

Você também pode gostar