Figure 1: CAS Mathematical Modeling

Size: px
Start display at page:

Download "Figure 1: CAS Mathematical Modeling"

Transcription

1 MODELING OF FLIGHT CONSTRAINTS FOR UNMANNED AERIAL VEHICLES (UAVS) BASED ON ONBOARD COLLISION AVOIDANCE SENSORS, SUPPORTED BY JAUS-BASED CONTROL STATION FOR MULTIPLE UAV OPERATIONS By TAN HAN YONG, OI TZE LIANG, TEOH WEI LIT, A/PROF GERARD LENG (DSO NUS COOPERATIVE SYSTEMS LABORATORY) ABSTRACT: Recently, NASA Ames and Dryden Flight Research Center conducted flight test for the multiple UAVs operation. The boids algorithms, which prevent collision between UAVs when flocking together, were tested. In the first section of the paper, we will extend the scope of the collision avoidance through the use of onboard sensors coupled with GPS. First, the paper will discuss the relation of the sensors range to the speed of the UAVs and how the safety factor, lag time and bank angle of the UAV affects this relationship. Secondly, the results of the relationship will be used to determine the constraints for inter UAV collision avoidance using sensor range detection for multiple UAVs operations. Data down/uplinks will be facilitated by the use of a Ground Control Station, which is based on the Joint Architecture for Unmanned Systems (JAUS) sponsored by the United States Department of Defence. JAUS is an upper level design, for the interfaces within the domain of Unmanned Ground Vehicles. It defines messages and component behaviors that are independent of technology, computer hardware, operator use, and vehicle platforms and isolated from mission. In the second section of the paper, an expository study is made on how JAUS may be adopted and implemented for multiple UAV operations. Using standard Internet-based computer network protocols, and Commercial- Off-The-Shelf hardware components, a Ground Control System (GCS) will be developed & discussed to deduce the implications of JAUS with respect to multiple UAV operations. 1. INTRODUCTION Any object (other UAVs or stationary objects) that is lying in the path of the UAVs is considered as an obstacle to the UAVs. The Collision Avoidance System (CAS) will help the UAVs to avoid collision with their obstacles. In [1], NASA Ames and Dryden Flight Research Center conducted test flight with two 15-ft wingspan UAVs using the boids algorithms [1]. These boids algorithms [1] ensure the UAVs flock together and also prevent collision between each other. [2] Boids CAS was implemented through a central computer station, which all the UAVs communicated to by sending their position given by their onboard electronics. The central computer will then re-direct the UAVs into new flight path to avoid collision between other UAVs or obstacles that are digitally implanted in the central computer.[1] In order to extend the scope of the CAS to enable detection of real obstacles placed in the environment with positions initially unknown to both the central computer and the UAVs, onboard proximity sensors will be added to the UAV configuration. Page 1 of 19

2 Also, with the current interest in using multiple inexpensive UAVs in unknown, cluttered and possibly hazardous environment to achieve a common goal, it is also desirable for the CAS to prevent collision between UAVs through the use of a central computer station as in the Boids CAS [1]. Thus it is desirable to design cooperative systems using a highly expandable architecture as the number of UAVs used increases. This must not just be able to expand in terms of number of platforms and capabilities, but also be able to easily accommodate new technological developments, and bridge across vendors and schemas. The Joint Architecture for Unmanned Systems (JAUS) presents itself as a desirable framework for such systems design of cooperative teams of UAVs. Hence, the main aim of this paper is to determine the flight constraints of UAVs taking into consideration the addition of onboard proximity sensors and effect on the number of UAVs that can be deployed with the implementation of the JAUS framework. 2. COLLISION AVOIDANCE SYSTEM (CAS) 2.1. MATHEMATICAL MODELLING OF CAS The model for the CAS to avoid any kind of obstacles is shown in Figure 1. The head-on collision between the UAV and the obstacle will be considered, as this is the worst-case scenario. R obs D R Vτ V Figure 1: CAS Mathematical Modeling A no-fly zone which is assumed to be spherical in shape is set around the obstacle with a no-fly zone radius R obs. When the UAV is moving towards the no-fly zone at a speed V, the collision avoidance onboard sensors detect the obstacle at a detection distance D. There is a short reaction lag time τ before the UAV executes a uniform circular turn of radius R to avoid the obstacle. Since the UAV is executing a uniform circular turn, the radial acceleration a r is given by the following equation. Page 2 of 19

3 V = Equation 1 R a r 2 Moreover, if we let D = k R obs, where k is the safety factor (k > 1), we can relate the detection range D to the radial acceleration a r, lag time τ and the speed V of the platform by Equation 2 k D = [ V + ( a τ ) (2 τ ) ( τ ) ] 2 rk V + V V + ark V + ar Equation 2 a ( k 1) r The radial acceleration a r of the UAV which is performing a banking maneuver in order to execute a uniform circular turn is given by Equation 3, where n is related to the bank angle θ by Equation 4. This results in Equation 5. 2 a r = g( n - 1) Equation 3 1 n = cosϑ Equation 4 a r = g tanθ Equation STATIONARY OBSTACLE COLLISION AVOIDANCE SYSTEM From Equation 2, we are able to draw the relation between the sensors detection range to the speed of the UAVs at different lag time and the no fly zone between the UAVs or stationary obstacles (otherwise known as the safety distance) in order to avoid collision. 3 different examples of proximity sensor range readings of 15m, 3, and 5m will be included in the following plots, illustrating the flight constraints limited by proximity sensor ranges. Relation of Detection Range to UAVs Maximum Speed at Different Lag Times The lag time τ is dependent on the UAVs size and the response time of the overall sensors / actuators system used by the UAVs. Figure 2 illustrates the relations between the sensors detection range to the UAVs maximum speed for different τ ranging from 1 to 4 seconds. Safety factor k is set at 3 and evasive banking maneuver is carried out with a 45. o bank angle (a r = 1.g) level turn. Page 3 of 19

4 D (m) D (m) vs V(m/s) (k = 3, θ = 45. o (a r = 1.g), level turn) τ = 4 τ = 3 Sensor Range = 5m X τ = 2 τ = 1 Sensor Range = 3m 25 2 Sensor Range = 15m V (m/s) Figure 2: Effects of Time Lag, τ on Detection Range and UAV's maximum speed Relation of Detection Range to UAVs Maximum Speed at different Safety factor, k. Figure 3 illustrates the relation between the sensors detection range to the UAV s maximum speed for different safety factors, k ranging from 2 to 5. The lag time is set at τ = 2s, and evasive banking maneuver is carried out with a 45. o bank angle (a r = 1.g) level turn. D (m) D (m) vs V(m/s) (τ = 2, θ = 45. o (a r = 1.g), level turn) Sensor Range = 5m Sensor Range = 3m Sensor Range = 15m k = 2 k = 3 k = 4 k = 5 5 V (m/s) Figure 3: Effects of Safety Factor, k on Detection Range and UAV s maximum speed Page 4 of 19

5 Relation of Detection Range to UAVs Maximum Speed for different bank angles, θ Figure 4 illustrates the relation between then sensors detection range to the UAV s maximum speed for different evasive maneuvers of level turns at banking angles of 26.6 o, 45. o, and 56.3 o and 63.4 o. These values correspond to radial acceleration, a r, of.5g, 1.g, 1.5g and 2.g respectively. The lag time is set at τ = 2s, and safety factor, k = 3. D (m) D (m) vs V(m/s) (τ = 2, k = 3) Sensor Range = 5m Sensor Range = 3m Sensor Range = 15m θ = 26.6 o θ = 45. o θ = 56.3 o θ = 63.4 o 5 V (m/s) Figure 4: Effects of radial acceleration, a r on Detection Range and UAV s Maximum speed Figure 2, 3 and 4 illustrate the relationship between sensor range detection distance and the maximum speed of the UAV at different lag times, safety factors and bank angles respectively. With reference to Figure 2, the X on the chart denotes the maximum speed a Searcher (IAI, max speed 198kph = 55m/s) UAV mounted with a Laser Technology Impulse 2 LR laser range finder (max range 575m) can fly at. With a sensor range of up to 5m, Searcher can fly at a top speed of close to 52m/s. This is under the assumption that the UAV system and maneuvering response lag time, τ is 3s, carrying out an evasive banking maneuver with a 45. o bank angle (a r = 1.g) level turn and the safety factor, k, set at 3. Figure 3 and 4 can be used in the same manner to determine other flight constraints at different safety factors and different bank angles. Hence, the charts can be used to determine flight constraints and operational limits and can be extended to gauge other different configurations provided information such as the bank angle and lag times are furnished. Page 5 of 19

6 2.3. INTER UAV COLLISION AVOIDANCE SYSTEM Collision avoidance between multiple UAVs can be achieved by having all UAVs communicate their position via onboard electronics to the central computer. However, the configuration could attain a higher level of robustness if it could still avoid collision with another UAV using onboard proximity sensors in the event of a communication breakdown between multiple UAVs and the central computer. Hence, in order to find the relationship between detection distance D of the UAVs to avoid each other and the maximum speed V of each platform, Equation 2 will be used. However, V in Equation 2 will be replaced by 2V since the relative speed of one UAV with respect to the other is 2V.This results in Equation 6. k 2 2 D = [2V + ( a τ ) 4 (4 τ ) ( τ ) 2 rk V + V V + ark V + ar a ( k 1) r 2 2 ] Equation 6 Collision avoidance between UAVs using sensor range detection serves as a backup to support a multiple UAV configuration, in case of a communications breakdown between the UAVs and the central computer which plans inter UAV collision avoidance. As such, the sensors selected should primarily be for avoiding stationary obstacles. Hence, the same 3 example proximity sensor range readings as provided in Figure 2, 3 and 4 would be selected and shown in the following charts. This would clearly indicate how using sensor range detection to provide inter UAV collision avoidance severely limits the flight constraints and operational range of the UAV. Relation of Detection Range to UAVs Maximum Speed at Different Lag Time D (m) D (m) vs V(m/s) (k = 3, θ = 45. o (a r = 1.g), level turn) Sensor Range = 5m X Sensor Range = 3m Sensor Range = 15m Figure 5: Effects of Time Lag, τ on Detection Range and UAV's Maximum speed for moving obstacles Page 6 of 19 τ = 4 τ = 3 τ = 2 τ = 1 V (m/s)

7 Relation of Detection Range to UAVs Maximum Speed at different Safety factor, k. D (m) D (m) vs V (m/s) (τ = 2, θ = 45. o (a r = 1.g), level turn) Sensor Range = 5m Sensor Range = 3m Sensor Range = 15m k = 2 k = 3 k = 4 k = 5 V (m/s) Figure 6 : Effects of Safety Factor, k on Detection Range and UAV s Maximum speed for moving obstacles Relation of Detection Range to UAVs Maximum Speed for different bank angles, θ D (m) Sensor Range = 5m D (m) vs V(m/s) (τ = 2, k = 3) Page 7 of 19 Sensor Range = 3m Sensor Range = 15m V (m/s) Figure 7 : Effects of radial acceleration, a r on Detection Range and UAV s Maximum speed for moving obstacles θ = 26.6 o θ = 45. o θ = 56.3 o θ = 63.4 o

8 Figure 5, 6 and 7 illustrate the relationship between sensor range detection distance to the maximum speed of the UAV for inter UAV collision avoidance using sensor range detection at different lag times, safety factors and bank angles respectively. As this is a backup to the multiple UAV CAS provided by the central computer, X on Figure 5 denotes the maximum speed the same Searcher UAV configuration as referred to earlier can now fly at. As seen from the chart, the maximum speed has decrease from 52m/s in Figure 2 to 26m/s in Figure 5. Similar trends can be observed from Figure 6 and 7 for the different safety factors and different bank angles. This shows that for using the same proximity sensor, once the backup CAS for multiple UAV operation is activated, the flight constraints are more severely limited than if the CAS is working only to avoid stationary obstacles. Hence, proximity sensor selection is critical if CAS with sensor range detection is to provide collision avoidance for inter UAVs as well as stationary obstacles. The next section will discuss on the limitations that the JAUS based Ground Control Station will have on multiple UAV operations. 3. JAUS-BASED CONTROL STATION FOR MULTIPLE UAV OPERATIONS As explained in the beginning of this paper, a Ground Control Station (GCS) can be useful in CAS operations. Therefore, to allow greater options for future CAS strategies, a GCS system is conceived, based on JAUS architecture INTRODUCTION TO THE JOINT ARCHITECTURE FOR UNMANNED SYSTEMS The Joint Architecture for Unmanned Systems (JAUS) is being developed in conjunction with the United States Department of Defense and many other members of industry and academia for use in research, development and acquisition of unmanned systems. The current version of the document [3] is divided into three large volumes; but it is with Volume II that we are most interested in because it concerns the design of the nodes and components being developed for the UAV system The main purpose of the Reference Architecture (RA) is to describe all functions and messages that shall be used to design new components. Joint Architecture for Unmanned Systems defines components for all classifications of Unmanned Systems, from tele-operations to autonomous. As a particular system evolves, the architecture is already in place to support more-advanced capabilities. To meet this requirement, four technical constraints are imposed on JAUS: Platform independence: Since unmanned systems will be based on a variety of missions, no assumptions are made regarding the vehicle platform. Mission isolation: Joint Architecture for Unmanned Systems defines a mission as the ability to gather information about or to alter the state of the environment in which the platform is operating. This allows the developer to construct the system to support a variety of missions. Page 8 of 19

9 Computer hardware independence: Advances in computer technology over the past couple of decades have seen rapid growth. The JAUS computer hardware constraint was put in place to ensure software and hardware portability as new systems are developed in the future. Technology independence: This constraint is similar to hardware independence, but focuses more on technical approach [7, 8]. In this particular application, the architecture makes no assumption regarding the method used to obtain joint positions. For example different manipulators could be outfitted by different position sensors that include encoders, potentiometers, or rotational variable differential transformers. Within these constraints, there is still much freedom for engineers to design their systems based on current technology. The technology-independent nature of the system makes the system perfect as a foundation for a scalable cooperative UAV system. With that in mind, the current basic architecture of the multiple UAV system is presented, based on the JAUS Reference Architecture template, in Figure 8. In this model, all UAVs and the GCS is considered to belong in the base squadron (Squadron ), which comprises the following main components Subsystem Commander (ID32) the GCS computer serves as this Communicator (ID35) The Wireless router serves as this Global Waypoint Driver (ID45) The PC/14 onboard the UAV serves to interprete instructions and translate into native Autopilot commands. More nodes will be added as the subsystem is upgraded for improved capabilities (ground vehicle pathfinders, lightweight operator viewer, etc). Another study on the use of JAUS for a robotic manipulator node is found in [4], to illustrate the benefits of JAUS to extend the capabilities of future unmanned systems. Page 9 of 19

10 SYSTEM Cooperative System SUBSYSTEM Squadron # Bridging Router Squadron #1 Ground Control Station (Subsystem Commander ID32) Wireless Router (Communicator ID35) NODE UAV #1 (ID 45) UAV #2 (ID45) Non-JAUS component Non-JAUS component Non-JAUS component UAV #i (ID45) Figure 8: System Architecture in JAUS Domain Model format Page 1 of 19

11 3.2. JAUS MESSAGE MODEL Currently, JAUS has been implemented on ground vehicles only, even though it is mandated for use for any unmanned system whether land, sea or air. In this project, JAUS will be used to define messages being passed for both ground and air vehicles to be commanded by the Control Station being delivered. Therefore, the following message classes apply a) Command class Commands sent from Control Station to platform(s) b) Query class Enquires states of platforms(s) c) Inform class Responds to Query messages d) Event Setup class Defines asynchronous events on Platform e) Event Notification class - Messages response to events defined by Event Setup. f) Node Management class for future use in more complex JAUS systems. g) User-defined class Covers all user-defined messages not defined in previous classes For the purposes of this project, the following commands are found to be required as additional for mission purposes. These are defined as both Event Setup (to define the event) and the Event Notification (message released when event occurs) classes Collision Avoidance Emergency event when Collision Avoidance System (CAS) sensors detect imminent collision with obstacle. Target Marked event when Target is marked by Automatic Targeting System (ATS) Virtual Collision event when simulated platform has collided in virtual space with real platform As the subsystem assumes more and more abilities, as a result of addition of nodes / components, the number of user-defined messages will probably increase. A complete subset of the JAUS Message set as defined in Volume II of the JAUS specification is documented in [5]. Page 11 of 19

12 3.3. BANDWIDTH REQUIREMENTS OF JAUS JAUS messages are of variable length. The currently defined header format defines a maximum message size of 48 bytes! However, most messages are expected to be of smaller size. To obtain the bandwidth requirements, we will assume the following conditions 1) Communications are being broadcasted from GCS to ALL UAVs 2) A proportion of messages are broadcasted by a UAV to all OTHER UAVs, in addition to GCS. It is known that for IEEE 82.11a/b/g [6,7,8], the frame overhead is 72 bytes, while for IEEE Bluetooth [9], the overhead is 16 bytes. Therefore we can obtain the byte rate for data transmission as follows. (Byte Rate) GCS = N [ NAC(S R + S O ) + BD(S C +S O )] - Equation 7 downlink uplink where N = Number of UAVs A = No. of JAUS reports per second B = No. of JAUS commands per second C = No. of packets required per JAUS reports D = No. of packets required per JAUS commands S R S C S O = Message Size of JAUS reports FROM UAV = Message Size of JAUS commands TO UAV = MAC Frame overhead for network protocol = 72 bytes (IEEE82.11x) or 16 bytes (Bluetooth) Now, C and D are given by the following formula, where C = RUP(S R / 146), - Equation 8 D = RUP(S C / 146) - Equation 9 Therefore, for the example of a system with JAUS commands at maximum size = 48 bytes, the upper limit of bandwidth requirements is given by (Byte Rate) GCS = N [ 3NA(S R + S O ) + 3B(S C +S O )] - Equation 1 Page 12 of 19

13 3.4. DISCUSSION OF THEORETICAL COOPERATIVE SYSTEM SCALING WITH RESPECT TO CURRENT WIRELESS TECHNOLOGIES We will base our discussion with regards to current mobile wireless technologies, as depicted in Table 1. We then define mobile wireless as systems which has built-in provisions for nodes to be roam across different coverage cells. This would then disqualify wired Ethernet, as well as Fixed Wireless Broadband (such as the IEEE WiMAX Standard ) technologies from consideration. Network Type IEEE 82.11b IEEE 82.11g IEEE Bluetooth Nominal bandwidth (Mbps) Frequency (GHz) Maximum IDA allowable transmission power (mw) Modulation DSSS OFDM FHSS Off-the-shelf Access Point Radio Range (m) - Reference [9,1] Table 1: Survey of Mobile Wireless Technologies According to Volume II, the maximum permissible size of a single JAUS message is 48 bytes. This is unfortunately, much larger than the maximum payload size of 146 bytes for a IEEE 82.1p data packet (PDU) [6], a limit set by the original Ethernet (IEEE 82.3) wired package. Any IEEE 82.x wireless network, which needs to be routed through the Ethernet, regardless of native implementation, will have this major limitation. It is known that large fragmentation of PDUs, due to need to break up large payloads, lead to degradation of transfer rates, as well as poor timing of message sending for packet-based protocols such as UDP. This is especially for the bandwidth-limited wireless networks. This implies that very large message sizes are to be avoided whenever possible. However, small message sizes would mean a very large header size to actual data size, which degrades the ratio of actual data sent per unit time. For the purpose of analysis, we shall consider 2 limits of JAUS message sizes, corresponding to WiFi Frame size limits: 146 bytes and 48 bytes. For this study, we also assume a command refresh rate of 1 Hz, and a reporting rate of 5 Hz. These values used in this study are valid for the following reasons a. Military wireless computer networks are also using the same IEEE 82.11x standards, as shown in reference [11]. The major difference lies in the much more sophisticated encryption technologies used. b. The maximum size of JAUS messages will be easily reached if military encryption algorithms are implemented in the message sending. Page 13 of 19

14 c. Otherwise, JAUS implementations should focus on delivering up to the maximum frame payload size of 146, so as to maximize payload to frame size ratio, for maximum network delivery efficiency d. The UAV(s) are to be semi-autonomous, so as to compensate for possible sporadic data link losses due to radio interference, loss of lineof-sight, weather etc. Hence the lower command data rate. e. State reports are to be regular, but set to also minimize overwhelming the shared network data link. From Figure 9, we find that a Bluetooth network may only support between 2-4 UAVs. This may be sufficient for deploying a basic system to test basic UAV functionality and small scale indoor deployment in an academic setting. Although limited in scalability, the Bluetooth system is, by design, actually very well suited for mobile applications. Unlike other networking systems, Bluetooth adapters are out-ofthe-box able to function both as relay stations, allowing the formation of scatternets. Such a system allows a cooperative system to be able to function in heavily cluttered settings where multi-path radio propagation predominates. The following reference [12] have more information concerning application of Bluetooth for urban UAV operations. 65 Network Bandwidth Requirements vs No. of UAVs - Bluetooth IEEE Network Bandwidth (kb/s) JAUS Msg Size 48 Bytes 146 Bytes Bluetooth Bandwidth 125kB/s No. of UAVs Figure 9: Graph showing effects of JAUS message sizes on low bandwidth system scalability Page 14 of 19

15 Network Bandwidth (kb/s) Network Bandwidth Requirements vs No. of UAVs - IEEE 82.11x WiFi systems IEEE 82.11a/g bandwidth JAUS Msg Size 48 Bytes 146 Bytes IEEE 82.11b bandwidth 1.38 MB/s6.75 MB/s No. of UAVs Figure 1: Graph showing effects of JAUS message sizes on higher bandwidth system scalability However, it is necessary to deploy WiFi-based systems in order to truly test cooperative algorithms. From Figure 1, a lower limit of 8 UAVs is possible with a bandwidth of 11 Mbps, while 17 UAVs may be controlled with a 54 mbps data link. An upper limit of about 3 UAVs is possible for a 54mbps network, if the maximum JAUS message size is limited to 146 bytes only DISCUSSION OF THEORETICAL COOPERATIVE SYSTEM NETWORK TOPOLOGIES WITH RESPECT TO CURRENT WIRELESS TECHNOLOGIES One point to note about all these network technologies are the short-range of the radio systems, due to legal restrictions by civilian telecommunications authorities, such as IDA. For military applications, radio signal amplifiers, in conjunction with suitable antennas, can be used to further extend IEEE 82.11x radio signals to far greater ranges than normal. To formulate the microwave propogation loss through free space, the following basic formula applies, as obtained from reference [13] L db = [2logd] + [2logF MHz] + k, where k = 32.4, if d is in km - Equation 11 Now, given the operating frequency of WiFi at ~24 MHz, we rewrite equation 11 as L db = 2logd Equation 12 Page 15 of 19

16 Setting our reference d =.3km, which is maximum range of DWL-22AP Wireless Access Point, we then calculate the reference loss, L db,,ref = 2log(.3) + 1 = Equation 13 Hence, K amplifier = L db L db,,ref = 2logd Equation 14 Introducing antenna gain, we obtain the final required system gain K system = 2logd K antenna - Equation 15 Equation 15 forms the basis for the Figures 11 and 12, where the antenna gain K antenna is obtained from the data sheets [15]. Based on military grade amplifier specifications [14], it is possible to amplify signals of C-band microwaves up to 25W. The 2.4GHz frequency used in IEEE 82.11b/g networks fall within this range. Therefore, the following figures apply, for purposes of comparison, using specifications from antennas available off-the-shelf from HyperLink Technologies as benchmark [15] Transmission Power (W) WiFi (11 Mbps) Amplifier Power vs Range Searcher LOS operational range Maximum WiFi Amplifier Power Searcher II LOS operational range Antenna Omnidirectional (5dBi) Yagi (14dBi) Panel Sector (2dBi) Parabolic Mesh (3 dbi) Transmission Range (km) Figure 11: Graph showing required WiFi Amplifier Power vs Transmission Range, for Amplified signal up to 25W Figure 11 shows the maximum line-of-sight range theoretically possible for operating UAVs with heavily-amplified WiFi signals. With a highly directional Parabolic Dish, it is possible to reach up to 2km ( well within the 1km Line-of-Sight (LOS) operating range of the SAF Searcher UAV), with a amplifier power of only 3W. Page 16 of 19

17 However, this also means that multiple directional antennas will be required to operate a multiple UAV system. The wider Field-of-View of the Panel Sector Antenna makes it more suitable for use as the operating antenna for the Ground Control Station, while still retaining sufficient gain to send signals within the Searcher s LOS range, at a power of about 7W. Inter-UAV WiFi (11 Mbps) Comms Power vs Range Amplifier Power (W) km 7 km 12 km Range (km) 1 watt max radio power UAV Antenna Type Figure 12: Graph showing required Inter-UAV WiFi Amplifier Power vs Transmission Range, for Amplified signal up to 1 W Omnidirectional (5 dbi) Yagi-Udo (14 dbi) Sector Panel (2 dbi) For discussion on inter-uav communications, we limit the maximum power to only 1W, due to power and weight limitations onboard small UAV systems. It is found that the range is severely limited, compared to GCS communication ranges. Therefore, an extensive relay system, or a scatternet ad-hoc network will need to be formed as part of the system architecture of the UAV system. Reference [16] is an example of an ad hoc network formed using both ground and UAV nodes, which could be used to extend the operating range of a cooperative system. 4. CONCLUSION In this paper, a mathematical model for collision avoidance system on UAVs has been developed. This model further is based on having onboard proximity sensors on the UAV and modeling the flight constraints taking into consideration the inputs from the sensors. As such, incorporating the Boids algorithm [1] and the improvements through the mathematical model developed, UAVs could possibly venture into unknown environment with real obstacles whose positions are initially unknown to both the UAV and the central computer. This model was also applied to provide collision avoidance between multiple UAVs, which brings to light the limitation the sensors will implicate on the flight constraints. Page 17 of 19

18 Additional flight constraints are also brought about by the implementation of an overall system architecture based on JAUS. JAUS sets a lower bound on bandwidth demanded of a mobile wireless computer network system. It is found that scalability up to 3 UAVs are possible with current off-the-shelf technology, although operational radius at this bandwidth is limited to about 1m due to transmitter power limits set by telecommunications regulations. Military-grade amplifiers with suitable antennas are theoretically able to extend this working range to match current Line-of-Sight capabilities of UAVs, allowing current UAVs the possibility to implement JAUS. Finally, for UAVs with limited transmission power, it will be necessary for the system architecture to support ad-hoc scatternet network topologies in order to allow extended range operations. 5. REFERENCES [1] NASA AMES, Testing for the Boids [2] Craig W. Reynolds, Flocks, Herds, and Schools: A Distributed Behavioral model, In Computer Graphics, 21(4), pp , July 1987 [3] Joint Architecture for Unmanned Systems (JAUS); Reference Architecture Specifications, Version 3.2, Volumes I III, JAUS Working Group, August 24, [4] Sosa O, Design & Implementation of a Modular Manipulator Architecture, Master s Thesis, Department of Mechanical & Aerospace Engineering, University of Florida, 24. [5] Oi T.L., Tan H.Y., Teoh W.L., Appendix: Preliminary JAUS Command Set for UAV search and mark mission, In Cooperative Search & Mark Report Set 2, DSO National Laboratories / National University of Singapore 26 [6] IEEE Std 82.11b-1999(R23), Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: Higher-Speed Physical Layer Extensions in the 2.4GHz Band, LAN/MAN Standards Committee IEEE Computer Society, 23 [7] IEEE Std 82.11g-23, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications; Amendment 4: Further Higher Data Rat Extensions in the 2.4GHz Band, LAN/MAN Standards Committee IEEE Computer Society, 23 [8] IEEE Std , Wireless Medium Access Control (MAC) and Physical Layer (PHY) specifications for wireless personal area networks (WPANs),LAN/MAN Standards Committee IEEE Computer Society, 25 [9] DWL-22AP AirPremier 82.11g / 2.4 GHz, Data Sheet, D-Link Systems Inc. 24 [1] DBT-12 PersonalAir Wireless USB Bluetooth Adapter, Data Sheet, D-Link Systems Inc, 24 Page 18 of 19

19 Systems Inc, 24 [11] HG-SL-21 SecNet 11 Plus PC Card, Data Sheet, Government Communications Systems Division (GCSD). Harris Corporation, 24 [12] Jae H.K, Lim M.S, Lim H.J, A Routing method for Control of Mobile Robot in Bluetooth Network, In International Technical Conference on Circuits/System, Computers and Communication, pp , Jul. 23. [13] Carr J.J., Practical Antenna Handbook, 4th Edition, Chapter 2, pp , McGraw-Hill Singapore, 21. [14] PA Series Power Amplifiers, for Airborne Telemetry, Datasheet, L3 Communications Telemetry-West, 24 [15] Datasheets for i) HG243G 3 dbi Reflector Grid WLAN Directional Antenna; ii) HG2415Y Randome Yagi WLAN Directional Antenna; iii) HG242P o Sector Panel WLAN Antenna; Hyperlink Technologies 26 [16] Brown, T. X., Argrow, B., Dixon, C., Doshi, S., Thekkekunnel, R.-G., and Henkel, D., Ad Hoc UAV Ground Network(AUGNet), In Proc. AIAA 3rd Unmanned Unlimited Technical Conference, Workshop and Exhibit, Chicago, Illinois, Sept. 24. Page 19 of 19