A U T O N O M O US R E N D E Z V O US, C O N T R O L A N D D O C K I N G E XP E R I M E N T REFLIGHT 2 Marco Barbetta (1)(a), A lessandro Boesso (2)(b), F rancesco B ranz (3)(c), A ndrea C ar ron (1)(d), Lorenzo O livieri (3)(e), Jacopo Prendin (4)(f), G abriele Rodeghiero (5)(g), F rancesco Sansone (3)(h), L ivia Savioli (3)(i), Fabio Spinello (1)(j), and A lessandro F rancesconi (6)(k)

Department of Information Engineering, University of Padova, Via G. Gradenigo 6/b, 35131 Padova, Italy (2) E S A-EAC, Linder Höhe, 51147 Köln, Ger m any, (3) &,6$6´*&RORPER´8QLYHUVLW\RI3DGRYD Via Venezia 15, 35131 Padova, Italy (4) Department of Computer Science, University of Venice, Castello 2737/b - 30122 Venice, Italy (5) Department of Physics and Astronomy, University of Padova, Vicolo Osservatorio, 3 35122 Padova, Italy (6) Department of Industrial Engineering, University of Padova, Via Venezia 1, 35131 Padova, Italy (1)

+39 3472291840, [email protected] (b) +39 349 4936095, boesso@gm ail.com (c) +39 049 827 6836, [email protected] (d) +39 049 827 7757, [email protected] (e) +39 049 827 6837, [email protected] (f) +39 329 7372714, nidnerp@gm ail.com (g) +39 333 6525360, [email protected] (h) +39 827 6836, [email protected] (i) +39 827 6835, [email protected] (j) +39 3491691998, [email protected] (k) +39 049 827 6839, [email protected] (a)

A BST R A C T This paper provides an overview of the ARCADE-R2 experiment, which is a technology demonstrator that aimed to prove the feasibility of small scale satellite and/or aircraft systems with automatic (i) attitude determination, (ii) control and (iii) docking capabilities. The development of such capabilities is fundamental to create, in the near future, fleets of cooperative, autonomous unmanned aerial vehicles for mapping, surveillance, inspection and remote observation of hazardous environments; small-class satellites could also take benefit from the employment of docking systems to extend and reconfigure their mission profiles. The experiment was composed by a supporting structure, which holds a small vehicle with one translational and one rotational degree of freedom, and its fixed target. The two units feature relative navigation sensors, attitude control actuators and a docking mechanism, along with pressure and temperature sensors, and wind probes. The experiment flew on board the BEXUS 17 stratospheric balloon on October 10th, 2013, where several navigation-control-docking sequences were executed and data on the external pressure, temperature, wind speed and direction were collected, characterizing the atmospheric loads applied to the vehicle. This paper describes the critical components of ARCADE-R2 as well as the main results obtained from the balloon flight.

The 4S Symposium 2014 ± Author initial and family name

1

1

INTRO DU C TIO N

!"##$%&$'()*+,-(Inserita bibliografia (ma senza rif incrociati)

In the last decade there has been a considerable interest in the development of highly capable smallscale vehicles for both space and aeronautic applications. On one hand, as regards the space field, the employment of miniature spacecraft with advanced capabilities would improve the execution of several new mission scenarios, such as I) on-orbit automated inspection and servicing of larger satellites, in order to extend their operational lifetime, replace their payloads or perform orbital manoeuvres, II) on-orbit assembly of low-cost large modular spacecraft, that would significantly benefit from standardization and miniaturization of the used modules, as well as from the possibility of launching them as piggybacks, III) active debris capture and removal, and IV) creation of distributed sensing instruments, employing closeformation clusters of inter-communicating nano-satellites. On the other hand, advanced cooperative Micro Aerial Vehicles (MAVs) could represent key technologies to be employed in several atmospheric applications, such as I) remote observation of hazardous or otherwise unreachable areas, i.e. toxic regions, urban canyons or interior of buildings, II) evaluation, rescue and surveillance operations and III) terrain mapping by means of distributed and cooperative fleets such vehicles, for both scientific or military purposes. In most of the current demonstrative applications, regarding both space and aerial fields, small spacecraft or vehicles are usually supported by a larger parent, which is dedicated to communication management, coordination, and refuelling and/or refurbishment operations. In this scenario, proximity autonomous navigation and attitude control subsystems, together with docking devices, represent the main qualifying technologies but, while they have already been developed for large scale spacecraft or manned vehicles, there is still a lack of applicable solutions for miniaturized automatic systems [3] [4] [5] [6]. The objective of this paper is to present the evolved version of the experiment ARCADE [7], named ARCADE-R2, which was developed at the Centre of Studies and Activities for SSDFH³CISAS G. &RORPER´ - University of Padova. ARCADE-R2 aimed to test autonomous navigation, attitude control and docking technologies in an extreme environment on-board a stratospheric balloon in the REXUS/BEXUS Program, in the wider framework of CISAS researches on small satellites technologies [8] [9]. 1.1 E xperiment O bjectives The objectives of the experiment ARCADE-R2 were divided in two classes: primary objectives and secondary objectives. Primary objectives dealt mainly with the determination of the subsystems performances according to the external environmental conditions; secondary objectives were defined as by-products of the experiment and the BEXUS flight. The primary objectives were: I. To test the custom-designed subsystems required to perform relative proximity navigation, relative attitude control and docking between a small aerial vehicle and its parent counterpart mounted on the BEXUS gondola; II. To evaluate the disturbances that affect control and docking operations at different altitudes during the ascent phase of the BEXUS balloon III. To relate the navigation, control and docking systems performances to the experienced disturbances; IV. To repeat several navigation-control-docking sequences in the low-disturbances environment expected in the float phase. The secondary objectives were: I. Determine the external temperature, pressure and density profiles during the whole flight; II. Determine the wind direction and velocity with respect to the gondola during the whole BEXUS flight.

The 4S Symposium 2014 ± Author initial and family name

2

1.2 T he R E X US / B E X US Programme REXUS/BEXUS is a yearly aerospace programme which offers to European university students the opportunity to carry out scientific experiments aboard sounding rockets (REXUS, Rocket-borne EXperiments for University Students) or stratospheric balloons (BEXUS, Balloon-borne EXperiments for University Students) and it realised under a bilateral Agency Agreement between the German Aerospace Center (DLR) and the Swedish National Space Board (SNSB). The Swedish share of the payload is been made available to students from other European countries through a collaboration with the European Space Agency (ESA). EuroLaunch, a cooperation between the Esrange Space Center of the Swedish Space Corporation (SSC) and the Mobile Rocket Base (MORABA) of DLR, is responsible for the campaign management and operations of the launch vehicles. Experts from ESA, SSC and DLR provide technical support to the student teams throughout the project [10]. In more details, the BEXUS programme consists in the launch of a stratospheric balloon with a volume up to 12 000 m3 (the dimensions are similar to those of a big hot-air balloon) to a maximum altitude of 35 km, depending on total experiment mass (ranging from 40 to 100 kg) and atmospheric conditions. The flight duration, considering both the ascending and descending phases, is 2-5 hours. During the mission, as the altitude varies, the experiments are exposed to different temperature and pressure conditions, which can reach extreme values as low as -90°C and a few mbar. 2

SYST E M O V E R V I E W

2.1 E xperiment L ayout The experiment goal is to perform autonomous docking sequences in extreme environment conditions between a small external vehicle, called SMAV (SMAll Vehicle), and its parent unit, named PROXBOX (PROXimity BOX). The PROXBOX is mounted on the gondola, which is the main structure lifted by the BEXUS balloon and where all the experiments are placed. The SMAV, on the other hand, is supported by an external rigid structure, called STRUT (STRUcTure), which connects the small vehicle to the gondola. The SM A V represents the chaser in the experimental docking scenario. It hosts dedicated navigation, control and docking devices as well as an eight-cell battery that makes it a fully independent vehicle from the power point of view. It relies, however, on the PROXBOX flight computer for data processing. It provides two controlled degrees of freedom (DOF): a rotational one, around the vertical yaw axis, and a translational one, perpendicular to the gondola. All the remaining DOFs are fixed. The ST R U T houses the devices needed to translate the SMAV and safely connects the external vehicle to the gondola. In particular, the small vehicle is mounted on a mobile supporting interface (SMAV-STRUT interface), that allows to rotate and translate the SMAV while ensuring a reliable mechanical connection. The STRUT also hosts a set of sensors to sound the external environment. The PR O X B O X represent the target in the reproduced docking scenario and it is mounted on the gondola with a wall facing outwards. It contains the parent-vehicle docking interface and navigation transmitters, a dedicated battery pack, most of the experiment electronics and the main processing unit. It determines the data sampling, receives commands, sends telemetry/data to the ground station, provides data storage and manages all the devices and sensors of the experiment. 2.2 Docking Subsystem ARCADE-R2 docking port can be described as small-scale, central, gender-mate interfaces, based on the concept of the Soyuz and ATV docking systems, and it was realized through the employment of low-cost and commercial components. It consists in a soft docking magnetic mechanism in the

The 4S Symposium 2014 ± Author initial and family name

3

!"##$%&$'()./,-(In genere userei tempo verbale al passato..se VLHWHG¶DFFRUGRYDQQRFDPELDWL,WHPSLYHUEDOLLQTXHVWRSDUDJUDIRDO simple past

drogue able to capture the probe and in a hard locking system composed by three peripheral solenoids. Dynamical simulations demonstrated it features a good tolerance to misalignments from nominal docking conditions [7]. The ARCADE-R2 docking mechanism is visible in Figure 1, that shows the two mating interfaces forming the probe-drogue mechanism. The first interface is mounted in front of the SMAV and is totally passive. It presents a conical shape with a ferromagnetic tip mounted on a spring-damper to absorb contact forces during docking procedure. The drogue is attached on the PROXBOX external wall and presents a conical shape to match the probe. To perform docking, the probe tip is softly captured by an electromagnet and thanks to a micro linear actuator it is pulled to its final configuration where the structural connection is achieved by means of three locking solenoids. The gondola and the SMAV interfaces are equipped with spring-mounted disks, creating a compressive force on the structure able both to secure the solenoids hard docking and to push away the SMAV during separation.

!"##$%&$'().0,-(External wall of the PROXBOX (suona meglio secondo me) !"##$%&$'().1,-(Has a complementary conical shape to match WKHSUREHVRQRHQWUDPEHFRQLFKHPDXQDYHUVRO¶HVWHUQRHO¶DOWUD YHUVRO¶LQWHUQR Secondo me è più preciso specificare) !"##$%&$'().2,-(During the docking procedure !"##$%&$'().3,-(Two spring-mounted disks are placed on the front sides of the SMAV and the gondola; they allow both to create a compressive force that secure the solenoids in the hard docking configuration, and to push away the SMAV during the release phase. (¶VRORXQVXJJHULPHQWRSHUULPDQHJJLDUHODIUDVH

Figure 1. ARCADE docking system: the probe-drogue mechanism is composed by two interfaces actuated by electromagnetic actuators; three solenoids create the solid joint in hard-docking configuration 2.3

Control Subsystem

!"##$%&$'().4,-(Stessa considerazione fatta sopra: il software è già stato sviluppato e testato«LRPHWWHUHLWXWWLWHPSLSDVVDWL o al massimo il presente (il futuro non ha senso)

The ARCADE-R2 Control Subsystem main task was to actively control the yaw SMAV movements and thus permit the SMAV-PROXBOX alignment. A correct attitude, during the approach to the proximity box located on the gondola, was necessary in order to successfully complete the experiment. For this purpose two main kinds of electric actuators were used: one reaction wheel on board the SMAV and a backup solution, i.e. a brushless DC motor directly connecting the SMAV to the support plate. A state-space control was designed considering the SMAV angular position and velocity and the reaction wheel speed. In order to achieve a more accurate control, the disturbance

The 4S Symposium 2014 ± Author initial and family name

4

due to the gondola attitude, that could cause small gravitational torques due to SMAV centre of mass misalignment, were rejected outside the feedback loop. As a matter of fact, the inclinometer located inside the proximity box provided the necessary angle measurements and allowed predicting the consequent torques that were therefore feed-forward as a RW command. Hence, the RW were actuated before these external torques could affect the SMAV attitude. Also, the linear movement of the SMAV was not included in the control loop and does not affect the yaw angle control in any significant way. This was because the linear motor was intended to accelerate in a few centimetres and thereafter it let the SMAV approach the gondola with constant speed. The differential equation that governs SMAV's angular acceleration is this: Ɏ Ʌሷ ൌ ‰‡ ቆ•‹Ƚ ‫‹• כ‬ሺɅ െ Ʌ଴ െ Ɏሻ ൅ •‹Ⱦ ‫ ‹• כ‬ቀɅ ൅ Ʌ଴ െ ቁቇ ൅ ɒ୵ െ ୤ Ʌሶ ʹ ൅ ɒୟୡୡ •‹ሺɅ଴ ൅ Ʌ െ Ɏሻ ൅ ɒ୫

(1)

Where M [kg] is the mass of the SMAV, g[m/s2] is the gravity acceleration, misalignment, , , 0 is the centre of mass angular position of the SMAV, e is the centre of mass eccentricity, w is the wind disturbance torque, Vf is the viscous friction coefficient, acc is the linear acceleration disturbance torque, m is the input torque provide by the reaction wheel. From the previous differential equations was obtained the state space representation. First of all rewrote in this way the first equation: šଵሶ ൌ šଶ

(2)

ͳ Ɏ šଶሶ ൌ ሾ‰‡ ቆ•‹Ƚ ‫‹• כ‬ሺšଵ െ Ʌ଴ െ Ɏሻ ൅ •‹Ⱦ ‫ ‹• כ‬ቀšଵ ൅ Ʌ଴ െ ቁቇ ൅ ɒ୵ െ ୤ Ʌሶ ʹ ൅ ɒୟୡୡ •‹ሺɅ଴ ൅ šଵ െ Ɏሻ ൅ ɒ୫ ሿ

(3)

where šଵ ൌ Ʌ. With the feed-forward technique we compensated the non-linear components in the equations, so the model became linear and its matrix representation was: Ͳ ሶ ൤šͳ൨ ൌ ൥ Ͳ šʹሶ

ͳ Ͳ Ͳ ˆ൩ ቂšͳቃ ൅ ൥ͳ൩ ɒ୫ ൅ ൥ͳ൩ ɒ୵ െ šʹ šͳ › ൌ ሾͳ Ͳ ሿ ቂ ቃ šʹ

(4) (5)

that is our state space continuous representation.

The 4S Symposium 2014 ± Author initial and family name

5

The control techniques used were a state feedback and an integral controller. The former allowed to adjust the transient parameters (rise time, settling time and overshoot), while the former was to control the steady state parameter (feedback error). Using a feed-forward loop, the disturbances induced by roll and pitch misalignments were rejected using the position data from the inclinometer placed inside the gondola. The logic scheme of the control is shown in Figure 2.

Figure 2: Logic control scheme We implemented a backup solution for the SMAV yaw control that was realized by a manual tuned PID controller. 2.4

Navigation Subsystem

!"##$%&$'().5,-(Stessec onsiderazioni: scegliete tempo verbale: presente o passato (io direi passato)

The experiment is provided with a relative navigation sensor for the determination of the SMAV linear position and yaw rotation, with respect to the PROXBOX. The sensor is composed by a transmitter infrared LED on-board the PROXBOX and two infrared photodiode receivers mounted on the face of the SMAV containing the docking mechanism, at a relative distance of 120 mm (see Figure 3). The signals generated by the photodiodes V i (i=1,2) are modulated by the SMAV ± PROXBOX relative position and attitude and are described in Eq. 6-7. In particular, they depend on the power emitted by the LED P LE D , the LED-photodiode distance r i, the photodiode sensitive area A i, the photodiode photosensitivity k i, the angle of incidence Di, the angle Ȗi between the LED optical axis and the line joining the LED to the i -th photodiode, and the amplification gain of the receivers conditioning electronics e i. The term IȖ represents the ratio between the LED radiant intensity (Wsr-1) at a given Ȗ and the radiant intensity at Ȗ= 0. Both IȖ and g(r) were determined experimentally. For simplicity, we define C i = P LE D A i k i e i, which is a constant contribution for each channel that has to be determined experimentally. The system is mathematically described by Eq. 6-10. As the translational motion of the SMAV can occur only along the LED central axis direction, the parameters ri, Įi DQGȖi result from simple geometrical relations involving the SMAV distance U and yaw rotation \ relative to the PROXBOX. The values of d and į are known from the geometry of the SMAV. By combining Eq. 6-10, it is possible to express V1 and V2 as function of solely the SMAV distance ȡDQG\DZrotation \. $GHGLFDWHGVRIWZDUHGHWHUPLQHVȡ and \ from the measured photodiode signals using an iterative inversion algorithm. The solution is always found in less than 10 steps. The infrared LED is driven by a sinusoid carrier frequency of 10 kHz and the receivers use photodiode amplifiers with a narrow band-pass filter centred at 10 kHz, followed by a precision

The 4S Symposium 2014 ± Author initial and family name

6

envelope detector. This solution was adopted in order to reject both direct current light component and ambient noise. The gain of the receiver section is adjusted in order to match the signal dynamics due to distance and rotation modulation. More details and the description of an alternative algorithm based on look-up tables can be found in [11].

Figure 3. Elements of the IR sensor on the PROXBOX and on the SMAV (left) and geometrical parameters of the sensor model (right) ܸଵ ൌ ܲ௅ா஽ ‫ܣ‬ଵ ݇ଵ ݁ଵ ݂ሺߛଵ ሻ݃ሺ‫ݎ‬ଵ ሻ …‘• ߙଵ ൌ ‫ܥ‬ଵ ݂ሺߛଵ ሻ݃ሺ‫ݎ‬ଵ ሻ …‘• ߙଵ

(6)

ܸଶ ൌ ܲ௅ா஽ ‫ܣ‬ଶ ݇ଶ ݁ଶ ݂ሺߛଶ ሻ݃ሺ‫ݎ‬ଶ ሻ …‘• ߙଶ ൌ ‫ܥ‬ଶ ݂ሺߛଶ ሻ݃ሺ‫ݎ‬ଶ ሻ …‘• ߙଶ

(7)

‫ݎ‬ଵǡଶ ൌ ඥߩଶ ൅ ݀ ଶ െ ʹߩ݀ …‘•ሺ߰ ‫ߜ ט‬ሻ

(8)

݀ •‹ሺ߰ ‫ߜ ט‬ሻ ߛଵǡଶ ൌ –ƒିଵ ቈ ቉ ߩ െ ݀ …‘•ሺ߰ ‫ߜ ט‬ሻ

(9)

ߙଵǡଶ ൌ ߛଵ ‫߰ ט‬

(10)

2.5 E nvironment Sensors In order to fully characterize the external atmospheric environment in terms of pressure, temperature and wind velocity, ARCADE-R2 is equipped with wind, temperature and pressure sensors. In order to determine the wind direction and intensity in the horizontal plane a mechanical anemometer is employed (Nuova Ceva Automation ANTC-V1 and ANTC-D1). The device is composed by a wind velocity sensor (ANTC-V1) and a wind direction sensor (ANTC-D1); both of them are very compact and lightweight (0.25 ± 0.3 kg) and, therefore, are mechanically suitable for mounting on ARCADE-R2. Both sensors are mounted on the outer edge of the STRUT one at each side of the SMAV. This location is selected since it is the farthest from the balloon gondola, keeping the anemometers far from the turbulences that generate around the vehicle structures.

The 4S Symposium 2014 ± Author initial and family name

7

The external ambient temperature is measured by means of a PT100 thermal sensor (thermistor). It is mounted on the SMAV plastic fin in order to maximize the contact with the external air stream while minimizing the conductive heat flow from the experiment. Aiming to gather the information on the atmospheric density and altitude, an absolute pressure sensor completes the environmental sensor suite. The selected model is the Honeywell SSC DANN 030PAAA5 and it is located inside the PROXBOX to protect it from low temperatures. 2.6

Support H ardware

The scientific payload of the experiment, made of the aforementioned subsystems, is supported in order to survive, operate and communicate, by the electronic, power and software subsystems. The whole electronics system has been conceived keeping in mind the need to control many types of actuators, and to acquire different types of physical measurements. Therefore an industrial CANopen bus solution has been adopted. Most of the selected devices are CANopen compatible, and are queried by the main intelligence of the system, an x86-based PC104 single-board computer (SBC). The SBC board contains a real-time OS, which executes the control application described in Software section. All the data acquisition (environmental sensors, switches) and actuation (motors, transmitters, and latches) is performed through the on-board CANopen bus, writing or reading data from industrial automation modules. A PC104-compatible CANopen interface allows the SBC to access the bus. Analog sensor data and power status (currents, voltages) are sampled by industrial automation modules, which include sensor conditioning, A/D conversion and bus interfacing. Digital inputs/outputs are sampled/actuated by industrial automation modules, provided by optoisolated inputs and open-drain driver outputs. Brushless motors (DPM, TECNOTION UC3) are driven by a switch-mode integrated PWM motor driver with built-in feedback acquisition and control. This driver is interfaced with motor winding hall sensors and motor encoder to provide control feedback. The brushed linear actuator (L12-P) is driven by a FIRGELLI Linear Actuator Control Board. Control setpoint is transmitted in current mode (4-20 mA) through a DAC industrial automation module. Both the LED emitter on board of the PROXBOX are driven by an analog sine wave generator followed by an impressed-current driver. Most of the signal conditioning on board of the SMAV is done by the Photodiode receiver. It includes the same chip used for the MEMS gyroscope, which acquires the analog demodulated signals and make them available on the CANopen bus. Because of the mechanical structure of the experiment the power systems of PROXBOX and SMAV are isolated and totally independent. This is due to the high power requirements and the fact that the PROXBOX and the SMAV have to simulate separated vehicles. Lastly, this helps to avoid the presence of thick power cables through the rotary joint of the carriage. The power supply consists of EVT ER34615M Lithium Thionyl-Chloride battery packs. Both the SMAV and the PROXBOX power systems are equipped with current and voltage diagnostics in order to monitor the voltages of the battery packs and the intensity of drained current. This is done with the purpose of controlling the charge state, avoiding any dangerous situation and taking the best performance from the batteries. The PROXBOX has a power supply commutation system that excludes the battery pack in case of low battery voltage and connects the experiment to the GONDOLA power line. With this backup system, in case of serious damage to the battery packs, power is guaranteed to accomplish emergency operations. The current software design involves two segments: the flight segment, operating on the PC104,

The 4S Symposium 2014 ± Author initial and family name

8

!"##$%&$'().6,-(Was or is !"##$%&$'().+7,-(Was or is

8$9$&$'-(OS which

8$9$&$'-(is 8$9$&$'-(part

8$9$&$'-(¶ 8$9$&$'-(¶ 8$9$&$'-(¶

and the ground segment, connected to the previous one through the E-Link. The flight segment software autonomously manages the SMAV motion control, the data storage on a solid state memory and the connection with the ground segment through the E-Link system. The ground segment allows an additional manual control on the experiment operations and to retrieve information from the on-board sensors. QNX has been chosen as the real time on-board operative system. The chosen programming language for control flight software is ANSI C and Python. The ground station applications and graphical front-ends, however, are written in JAVA. 3

L A U N C H C A M PA I G N

A typical launch campaign for a BEXUS program lasts ten days. The first days are dedicated to mount the experiment in a flight configuration and to perform several tests in order to verify the functionality of all the experiments mounted on-board the same gondola. The launch windows are determined at the beginning of the campaign and daily updated. The launch campaign of ARCADE-R2 started October 4th and ended October 14th 2013. The experiment was launched on-board BEXUS-17 balloon October 10th 2013 at 17.15 UTC (day 6 of the campaign). The maximum altitude reached during the flight was about 27 km. At 18.46 UTC, when the balloon was still in the ascending phase, the E-Link connection was lost because of the extremely low temperatures reached inside the E-Link box, causing the stop of communication between ARCADE-R2 and the ground base. Despite the extreme external conditions, the data registered through the temperature sensors placed both in the SMAV and the PROXBOX confirmed that the temperatures were kept above -10°C until the cut-off. ARCADE-R2 softly landed in the Finnish forest at 00.01 UTC. It was verified that the electronics inside the PROXBOX continued to work until it was switched off by the recovery team around noon of October 11th. Main damages were registered only on the protection plate placed under the STRUCT, while the whole experiment was almost intact. 4

M ISSI O N R ESU L TS

In this section, the main results obtained during the flight operations are presented. 4.1 Docking Subsystem During the integration on the balloon gondola, ARCADE-R2 was continuously subjected to test. Data collected during such trials confirmed all the docking actuators and sensors functionality and permitted to check the perfect alignment between probe and drogue. Last, basic measurements were implemented to improve the docking automatic procedure, calculating the required actuation strokes for each motor. The subsystem was declared ready to flight and was launched in the mated configuration, in order to reduce the transient loads that could affect the motors, and was released at 18:20 to perform its programmed procedures. During the flight, ARCADE-R2 successfully performed three release operations and two docking procedures (Figure 4). In the first flight test, it was observed that the drogue actuator was not able to capture the probe, due to minimal thermal deformations in ARCADE-R2 (the temperature dropped of about 1.55°C every minute for the first 30 minutes): it was sufficient to increase the previously implemented actuator stroke of 1 mm to allow a better contact between probe and drogue and realize complete docking sequences. Unfortunately, after the link loss, no other procedures were realized, although ARCADE-R2 survived and continued to collect environmental data during the whole flight and after landing. Data collected from the experiment demonstrate its features and was enough to declare the total

The 4S Symposium 2014 ± Author initial and family name

9

!"##$%&$'().++,-(Scegliete che tempo verbale usare

!"##$%&$'().+/,-(Allowed to declare (was enough mi sembra più colloquiale)

success of the docking subsystem: the link loss did not influence the overall subsystem success. These results demonstrated the mechanism intrinsic robustness and confirmed the design process presented by Boesso [7]; future works may improve the docking interfaces to reduce the mass utilization and the power budget, evaluating the implementation of passive hard docking latches. Last, the whole design process and the data collected during the flight gave important suggestions to similar docking systems developed by CISAS in the framework of small satellites docking system development [8] [9].

Figure 4. ARCADE-R2 docking procedures during test (left) and flight (right); ARCADE successfully performed three releases and two docking procedures during the flight 4.2

Control Subsystem

The control subsystem mission results were divided between the various actuation, feedback sensors and control algorithm options available on ARCADE-R2. In detail: - UHDFWLRQZKHHOEDFNXSPRWRU -

PRWRUHQFRGHU,5QDYLJDWLRQVHQVRUJ\URVFRSHV

-

3,'FRQWUROOHU66FRQWUROOHURQO\IRUUHDFWLRQZKHHO

!"##$%&$'().+0,-(Format (distanza tra le righe)

The feedback loop on the backup motor was a PID position loop internal to the motor driver; it was not possible to control it with the custom controllers developed for ARCADE-R2. Unfortunately, a few days before launch the SMAV gyroscope experienced a major failure. In addition, the SS feedback loop on the IR sensor signal showed to be unreliable during pre-flight tests and, consequently, was not included within the flight operations. The backup motor positioning always worked properly. Several homing and pointing manoeuvres were executed during the flight and the accuracy was always within the target performances. Figure XX shows the results of an example manoeuvre.

The 4S Symposium 2014 ± Author initial and family name

10

Figure 5: SMAV pointing manoeuvre: backup motor on encoder feedback The reaction wheel control with encoder feedback worked properly during the two manoeuvres tested, one with PID controller and one with SS controller. A 4 degrees dead band was implemented in order to avoid the saturation of the RW. Figure 5 shows the pointing angle with RW controller, encoder feedback and PID controller, while Figure 6 shows the pointing with SS controller. Some manoeuvres were tried on the IR sensor feedback but the results were not satisfying. Figure 8 shows the yaw angle vs. time for one of these manoeuvres.

Figure 6: SMAV pointing manoeuvre reaction wheel on encoder feedback, PID controller

The 4S Symposium 2014 ± Author initial and family name

11

Figure 7: SMAV pointing manoeuvre: reaction wheel on encoder feedback, SS controller

Figure 8: SMAV pointing manoeuvre: reaction wheel on IR sensor feedback, PID controller 4.3 Navigation Subsystem The infrared navigation sensor was tested during each automatic procedure that was executed by the experiment. At the beginning of each of them, the calibration constants of the photodiode receiver acquisition channels, C 1 and C 2, were determined automatically. This solution was adopted as both the LED emitter and the photodiode receivers are sensitive to temperature variations, requiring a periodic calibration during the whole flight. Then, the reconstruction algorithm was applied to the measured photodiode voltages, providing a real-WLPHHVWLPDWLRQRIWKH60$9GLVWDQFHȡDQG\DZ URWDWLRQȥUHODWLYHWR the PROXBOX. Figure 9 shows the results of the estimation process compared to the values of SMAV position and attitude provided by linear and rotary encoders, and indicates the instants when the calibration constants were determined.

The 4S Symposium 2014 ± Author initial and family name

12

Figure 9. Estimates of the SMAV linear position (up) and yaw rotation (bottom) compared to the reference values given by the linear and rotary encoders. The dotted lines indicate when the calibration constants were determined The relative navigation VHQVRUSHUIRUPDQFHLVH[SUHVVHGLQWHUPVRIVWDQGDUGGHYLDWLRQıȡıȥ of the GLIIHUHQFH EHWZHHQ WKH HVWLPDWHV ȡ ȥ DQG WKH UHIHUHQFH YDOXHV JLYHQ E\ WKH HQFRGHUV 7KH FDOFXODWHGYDOXHVRIıȡ DQGıȥ are, respectively, 0.017 m and 2.7 °. 4.4

E nvironment Sensors

The environmental sensors (external temperature, pressure, wind sensors) were able to collect data throughout the flight. External temperature conditions were harsh during BEXUS-17 flight and the ARCADE-R2 temperature sensor recorded data as low as -58°C. Since the sensor is mounted on the experiment structure, the measured temperature profile could be slightly different from the actual air temperature. The pressure sensor recorded a minimum atmospheric pressure of 10 mbar reached during the floating phase. The pressure time profile suggests that the actual value should not have fallen below the minimum sensor threshold, since the signal is never locked to a fixed constant value. Figure 10 shows the altitude profile as registered by E-Bass GPS sensor compared to the reconstructed altitude from ARCADE-R2 pressure data. Some discrepancy is present and it is probably due to the zero accuracy of ARCADE-R2 pressure sensor.

The 4S Symposium 2014 ± Author initial and family name

13

Figure 10: BEXUS-17 altitude profile The wind speed sensor was supposed to measure the horizontal wind intensity over time and worked properly throughout the flight. The sensor is affected by a minimum threshold speed (zero error) required to overcome the initial friction torque and to trigger the sensor rotation; this threshold is not constant with changing air density. The wind direction sensor worked properly. Nevertheless, when the measured wind speed is close to the sensor threshold, it is not possible to know if the measured wind direction is correct, since the aerodynamic forces may be too low. The environmental sensors provide the information required to determine the disturbance torques acting on the SMAV. The aerodynamic forces are strongly dependent on the wind speed, thus an estimate of the actual wind speed is required. The already mentioned wind sensor threshold is due to the initial friction at the sensor rotor. The threshold wind speed, vthr, relation as a function of air density is Eq. 11:

‫ݒ‬௧௛௥ ൌ ඨ

ߩ௚௡ௗ ‫ݒ‬ ߩ ௧௛௥̷௚௡ௗ

(11)

where ȡ is the air density, ȡgnd is the air density at ground level and vthr@gnd is the threshold wind speed at ground level. The wind speed considered for the calculation of the disturbance torque is the sum of the threshold speed and the measured speed above the threshold. Assuming always a lateral incidence of the wind against the experiment, the disturbance torque, Tdis, is estimated with Eq. 12: ͳ (12) ܶௗ௜௦ ൌ ߩ‫ ݒ‬ଶ ‫ܥ‬௣ ‫ܾܣ‬ ʹ where v is the wind speed, C p is the pressure coefficient (assumed to be 1), A is the SMAV fin surface area and b is the moment arm measured from the SMAV rotation axis and the fin center of pressure. The computed disturbance torque is presented in Figure 11.

The 4S Symposium 2014 ± Author initial and family name

14

Figure 11: aerodynamic torque on SMAV The frequency distribution of both the aerodynamic torque and wind direction has been estimated from the collected data. It appears that almost all the wind gusts are limited below 0.05 Hz, confirming that the design assumption of 1 Hz was quite conservative. The disturbance torque due to wind gusts is fairly low. During the design phase a worst case torque of 30 mNm was assumed and this value matches the peak torques estimated from flight data. Nevertheless, this peak value was experienced by the experiment only during the ascent and descent phases. The majority of the flight was characterized by torques below 5 mNm with peaks at 10 mNm. In addition, during the floating phase the disturbance torque was extremely low and hard to determine. Given these results, it is clear that the environmental disturbances were very low and, reasonably, did not affect the SMAV motion for the majority of the flight. During the ascent phase the disturbance was more considerable but definitely within the rejection capability of the control system. 4.5 Support H ardware Support hardware performed well during flight and thoroughly granted survival, operativity and communication to the experiment. The electronic system experienced no failures neither on data conditioning or actuator operation. The power system sustained the power needs of the experiment subsystem as expected, and the reserve power on the batteries let the SBC to remain alive until recovery, as mentioned. The software system experienced no failures or crashed, correctly delivered telemetry and grated telecommand access. Data logs were successfully extracted after flight and no missing record were suspected. 5

C O N C L USI O NS

ARCADE-R2 is a technology demonstrator that aimed to prove the feasibility of small scale satellite and/or aircraft systems with automatic (i) attitude determination, (ii) control and (iii) docking capabilities. The experiment consisted of a 2-DOF moving vehicle and a fixed assembly with a docking interface; it flew on board the BEXUS 17 stratospheric balloon on October 10th, 2013, where several navigation-control-docking sequences were executed and data on the external pressure, temperature, wind speed and direction were collected, characterizing the atmospheric loads applied to the vehicle. During the flight, ARCADE-R2 successfully performed three release operations and two docking

The 4S Symposium 2014 ± Author initial and family name

15

procedures. Data collected from the experiment demonstrate its features and was enough to declare the total success of the docking subsystem. These results demonstrated the mechanism intrinsic robustness. The infrared navigation sensor was tested during each automatic procedure executed by the experiment, providing real-time estimation of the vehicle distance and attitude with respect to the fixed assembly. The average accuracy of the estimates were, respectively, 17 mm and 2.7 °. The environmental sensors (external temperature, pressure, wind sensors) were able to collect data throughout the flight. [risultati del controllo]

:";#<&&$'-!"#$%&!'#()

The ARCADE-R2 experiment and the positive results from BEXUS-17 flight set an important baseline for future development in the field of docking and rendezvous technologies for small autonomous vehicles, and is currently a valuable technology asset in the CISAS (University of Padova) research activities.

6

R E F E R E N C ES

[1] -RKQVRQ1/DQG0F.QLJKW'6$UWLILFLDO6SDFH'HEULV2UELW%RRN&RPSDQ\0DODEDU )ORULGD [2] *UQ(HWDO&ROOLVLRQDO%DODQFHRIWKH0HWHRULWLF&RPSOH[,&$5869RO±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

The 4S Symposium 2014 ± Author initial and family name

16