Collaborative Control of UAV/UGV

Transcription

1 The 11th International Conferene on Ubiquitous Robots and Abient Intelligene (URAI 2014) Nov , 2014 at Double Tree Hotel by Hilton, Kuala Lupur, Malaysia Collaborative Control of UAV/UGV Jae-Keun Lee 1, Hahin Jung 1, Huosheng Hu 2 and Dong Hun Ki 3 1 Departent of Advaned Engineering, Kyungna University, Changwon, , Korea (E-ail: laughaway_@nate.o, zeroful1@naver.o) 2 Shool of Coputer Siene & Eletroni Engineering, University of Essex, Colhester, United Kingdo (E-ail: hhu@essex.a.uk) 3 Departent of Eletial Engineering, Kyungna University, Changwon, , Korea (E-ail: dhki@kyungna.a.kr) Abstrat In this study, a ooperative UAV and UGV platfor is proposed to obtain a wide range of visual inforation. The UAV reognizes a pattern arker on UGV and traks the UGV without user ontrol. It an provide wide range of visual inforation for a user in the UGV. The UGV by a user is ontrolled equipped with an aluinu board. And the UAV an take off and land on the UGV. The UAV uses two aeras; one aera is used to reognize a pattern arker and another is used to provide a wide range of visual inforation to the UGV's user. It is guaranteed that the proposed visual-based approah detets and traks the target arker on the UGV, and then lands well. The experiental results show that the proposed approah an effetively onstrut a ooperative UAV/UGV platfor for obtaining a wide range of vision inforation. Keywords UAV, UGV, Vision, Quadopter, Trajetory Following 1. Introdution Reently, UAV and UGV have been atively researhed. The UAV takes off fro the UGV, traks its loation, and lands at the UGV. In this situation, iage reognition is oonly used by the UAV for reognizing the UGV. A blak and white irle is positioned on the top side of the UGV and the UAV reognizes this irle while oving[1][2]. In another study, two LEDs are loated at top positions on the UGV and UAV; the distane between these two LEDs fro a wathing position is used for positional deterination[3]. However, previous studies[1-3] have suggested that blak olors and LEDs are ineffiient during the day; it is diffiult to keep trak of a blak irular objet when the sun appears or if the light shines on the UGV and UAV, ausing the disadvantage of iproper traking. In this study, a ooperative UAV and UGV platfor is proposed. The UAV is plaed in front of the field of view of a aera to ollet a wide range of inforation so the UGV s user an keep traking the UGV by the UAV. A UAV in the air is diffiult to observe fro the ground beause of liited sight. Therefore, UAVs are suitable for reonnaissane and surveillane issions. However, the short operating tie of UAVs offers a disadvantage when attepting to reah a destination: owing to the onstraints of tie, it is diffiult to onvey visual inforation to the UGV s user. To perfor the duties required for the user, the operating tie of oerially available UAVs is usually 20 to 25 inutes; this tie ay also be shortened. Therefore, in this study, to overoe these shortoings, the UAV is fixed on the UGV when no visual inforation is required. The UAV takes off and provides visual inforation when required by the UGV s user. After the end of the ission, the UAV lands on the UGV to harge. After harging, if the UGV s user requires a wide range of visual inforation, oands are given to the server PC for the UAV to take off. 2. Overall Syste The overall platfor of the proposed UAV/UGV is shown in Fig 1; this figure inludes inforation about the UAV, the UGV, and the server PC. The UGV is a obile robot with four wheels ontrolled by the user. The UAV onsists of a onverted DJI Phanto 2 frae. Norally, the UAV is loated on top of the aluinu board on the UGV, and the UGV s user gives the takeoff oands using the server PC. During takeoff, the botto aera of the UAV transfers an iage of the arker to the server PC. Siultaneously, the front of the aera provides the user with a wide range of visual inforation. The server PC reognizes the aera iage arkers and transits pakets of urrent loations to ove the UAV. Fig. 1. Overall Syste The ontrol struture of the overall syste is shown in Fig 2. The UAV is oposed of two iroontrollers. A Phanto 2 NAZA-M is ebedded in the iroontrollers. Gyro sensor, opass sensor, and GPS are used for hovering and attitude ontrol; a bus signal fro eah otor provides input for ontrolling the otors /14/$ IEEE 641

2 Fig. 2. Control Arhiteture 3. UGV-UAV 3.1 UGV The UGV is shown in Fig 3. The struture inludes four wheels, eah of whih are ontrolled by four respetive otors. The iroontroller for ouniation and PWM output uses a DSP, and a WiFi odule is used for ouniation between the UGV and the server PC. For takeoff and landing, the UAV is attahed to the aluinu plate on top of the ontrol board. The arker is loated in the enter of the plate. Aluinu plates allow easy takeoff and landing for the UAV; the aluinu board is larger than the UAV. The UGV is ontrolled by the user s oands fro the server PC. The UGV is shown in Fig 3. Paket oands fro the server PC are listed in Table 1. Table 1 Transission pakets of the UGV. Coand Start Bits Couniation paket Parity Bits Front FF AA Bak FF AA Left FF AA Right FF AA 07 4F Parity Bit = ~((0xFF+0xAA+(04~07))&0xff) Fig. 3. Coposition of UGV : Overall Coposition Internal Coposition Table 2 Speifiations of the UGV. Body Inluding aluinu board Length Width Height Weight 4.3 kg Atuation tie 90~100 in 3.2 UAV A DJI Phanto 2 is used for the frae and otors of the UAV. For autonoous traking, the UAV is reodeled and equipped with ontrollers and other devies, as shown in Fig 4. A WiFi odule is attahed for ouniation between the server PC and the iroontrollers. An IP aera is attahed at the botto to reognize the UGV s arker. For sooth ouniation of the IP aera iages, the IP aera and router ouniation are wired. The router is equipped on the botto of the UAV. Table 3 Speifiations of the UAV. Exept propeller Inlude propeller Length Width Height 180 Weight 2.6 kg Atuation tie 10~15 in 642

3 through oordinates. The spae angle, θ e, of the UAV and UGV an be obtained by subtrating 90 fro θ. θ = θ 90 (2) e s The top length of the arker, defined as M, is used to ontrol the height of the UAV's. In atual experients, the UAV aintains an approxiately 2 t height for sipler d traking. The arker length value, M, is defined at a distane of 2 between the arker and the UAV, aording to prior aera experients. The error, e z, is obtained by subtrating the length of the referene arker fro the length of the real iage arker. = d s e M M (3) z Fig. 4. Coposition of UAV : Front Coposition Rear Coposition 4. Traking Method The arker reognition ethod is used for reliable traking of the UAV. The arker reognition algorith is based on Visual Studio; the server PC uses the OpenCV library progra for proessing. Depending on the positions of the arkers for the four attitude ontrol values, the values are transitted via WiFi. The UAV is oved through four attitude ontrols: roll, pith, yaw, and height. As shown in Fig 5, position error, e, is obtained at the enter of the iage, along with the UAV Q at the enter of the pattern with the UGV M. M x Qx M = Q = M y Q y e x e= =M -Q e y As shown in Fig 5, the angle differene, θ, of the x-axis of the UGV is defined for the UAV and UGV to deterine the twisted angle and establish the overall piture (1) Fig. 5. UGV Traking with UAV Collaboration : Roll, Pith Yaw, Height The four paraeters, obtained earlier by using PD ontroller, Q, are alulated. 643

4 Q Q P + k e + k e ' r p x d x Q P + k e + k e ' p p y d y = = Q P h e h e ' h p z d z θ θ ' Qy P k k p e d e (4) Transission pakets of the UAV are listed in Table 4. Table 4 Transission pakets of the UAV. FF BB CH1 CH2 CH3 CH4 P Parity Start Bits 78~108 Bits Parity Bit = ~((0xFF+0xBB+CH1+CH2+CH3+CH4)&0xFF) 5. Result of Experient The proposed platfor is verified through experients with the produed UAV and UGV. Fig 6 shows a graph of eah UAV attitude ontrol value of roll, pith, and height during the UGV traking. Eah ontrol value is obtained through expression (5). E + ' r ke p x ke d x E = ke + ke' p p y d y ' Eh he he p z d z (5) Distane values for eah attitude ontrol error are as follows: roll is E =± 20[], pith is E =± 13[], and r p height is E = 0 ~ 18[]. During outdoor experients, h the effets of wind on the UAV inreased the error, but error-free traking an be identified in Fig 7. () Fig. 6. Eah traking value: Roll Pith () Height Fig. 7. Error value in traking Fig 7 shows an error value in traking. Error value is obtained by roll and pith during traking and through expression (6). E t E E (6) 2 2 = (E + ) t r p 644

5 Fig. 8. Experiental traking of UGV by UAV Fig 8 shows the experient traking the UGV by the UAV. Iages an be reeived of up to 30 fps, but the ouniation status is uneven, aintaining approxiately fps. The traking algorith for the sapling tie is set to 0.25 s. The proposed researh results in ollaborative ontrol of the UAV/UGV [4]. 6. Conlusion In this study, a platfor was proposed to operate a UGV and a UAV to obtain a wide range of visual inforation. As a result, the UAV traks the UGV and visual inforation is transitted to the UGV. This visual inforation is eployed by the UGV s user for avoiding obstales, prediting onditions, and planning driving paths. The single reonnaissane UAV is ineffiient for ontinuous reonnaissane beause of its short operating tie. However, these drawbaks are overoe through operation with the UGV. In addition, reliable traking is ade possible by using a arker pattern to ounterat the effets of external light and objets. The syste proposed in this study is expeted to be applied for speial vehiles requiring onidiretional paths, broadasting ars requiring onidiretional filing, and oand ary ars hanged to UGVs. Aknowledgeent This researh was supported by Basi Siene Researh Progra through the National Researh Foundation of Korea(NRF) funded by the Ministry of Eduation, Siene and Tehnology (2013R1A1A ). Referenes [1] Jae-Young Choi and Sung-Gaun Ki, Study on the Iproved Target Traking for the Collaborative Control of the UAV-UGV, Journal of Institute of Control, Robotis and Systes, Vol 19, No. 5, pp , [2] Cheng Hui, Chen Youheng, Li Xiaokun and Wong Wing Shing. Autonoous Takeoff, Traking and Landing of a UAV on a Moving UGV Using Onboard Monoular Vision, Chinese Control Conferene, pp , Jul [3] Wei Li, Tianguang Zhang and Kolja K uhnlenz A Vision-Guided Autonoous Quadrotor in An Air-Ground Multi-Robot Syste, International Conferene on Robotis and Autoation, pp , May [4] PROJECTS>Video Clips (Cooperative UAV/UGV Platfor for a Wide Range of Visual Inforation) 645