Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) - 07.26.17

Overview | Description | Applications | Operations | Results | Publications | Imagery

ISS Science for Everyone

Science Objectives for Everyone
Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) are bowling-ball sized spherical satellites. They will be used inside the space station to test a set of well-defined instructions for spacecraft performing autonomous rendezvous and docking maneuvers. Three free-flying spheres will fly within the cabin of the station, performing flight formations. Each satellite is self-contained with power, propulsion, computers and navigation equipment. The results are important for satellite servicing, vehicle assembly and formation flying spacecraft configurations.
Science Results for Everyone
Bowling in space! Those bowling balls were actually satellites used to test autonomous maneuvers. The investigation developed modular algorithms, on-board navigation, position and attitude estimators, fault detection, and a closed-loop delta velocity mixer using real-time measures to control thrusters. Tests included successful collision avoidance, recovery of two lost satellites, scatter maneuvers, trajectory tracking, realignment, coupling multiple satellites, path-planning algorithms, tele-operation of robots, and manual abort and navigation. Final integrated tests culminated with successful docking to fixed and tumbling targets, the latter a space first. Researchers concluded that tele-operated robotic spacecraft can conduct repairs, maintenance, inspections and monitoring, and de-orbit malfunctioning or defunct spacecraft.

The following content was provided by Alvar Saenz-Otero, Ph.D., and is maintained in a database by the ISS Program Science Office.
Experiment Details


Principal Investigator(s)
Alvar Saenz-Otero, Ph.D., Massachusetts Institute of Technology, Cambridge, MA, United States

Danilo Roascio , Ph.D., MIT, Cambridge, MA, United States
Jonathan P. How, Ph.D., Massachusetts Institute of Technology, Cambridge, MA, United States
Edward Wilson, Ph.D., Intellization Incorporated, Moffett Field, CA, United States
Gregory E. Chamitoff, Ph.D., Johnson Space Center, Houston, TX, United States

Defense Advanced Research Projects Agency, Washington, DC, United States
Payload Systems Incorporated, Cambridge, MA, United States
United States Department of Defense Space Test Program, Johnson Space Center, Houston, TX, United States

Sponsoring Space Agency
National Aeronautics and Space Administration (NASA)

Sponsoring Organization
National Laboratory - Department of Defense (NL-DoD)

Research Benefits
Information Pending

ISS Expedition Duration
October 2003 - April 2004; April 2006 - September 2012

Expeditions Assigned

Previous Missions
SPHERES is a continuing investigation on the ISS which began during Expedition 8. Testing with one satellite was performed early during expedition 13. The second satellite was delivered to ISS on STS-121 which allowed testing of the two satellite configuration beginning in August 2006. The third satellite was delivered on STS-116; the three satellite configuration testing began on ISS Expedition 14.

^ back to top

Experiment Description

Research Overview

  • SPHERES consists of three self-contained, free flying satellites for use inside the ISS.

  • SPHERES can test algorithms related to relative attitude control and station-keeping between satellites, re-targeting and image plane filling maneuvers, collision avoidance and fuel balancing algorithms, as well as an array of geometry estimators used in various missions.

  • The SPHERES facility acts as a free-flying platform that is capable of accommodating varying mounting features and mechanisms in order to test and examine the physical or mechanical properties of materials in microgravity.

  • Individual satellites communicate with each other and an ISS laptop through a low-power 900 MHz wireless link.

SPHERES is a testbed for formation flying by satellitesthe theories and calculations that coordinate the motion of multiple bodies maneuvering in microgravity. To achieve this inside the ISS cabin, bowling-ball-sized spheres perform various maneuvers (or protocols), with one to three spheres operating simultaneously. The Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) experiment tests relative attitude control and station-keeping between satellites, re-targeting and image plane filling maneuvers, collision avoidance and fuel balancing algorithms, and an array of geometry estimators used in various missions. SPHERES consists of three self-contained satellites, which are 18-sided polyhedrons that are 0.2 meter in diameter and weigh 3.5 kilograms. Each satellite contains an internal propulsion system, power, avionics, software, communications, and metrology subsystems. The propulsion system uses carbon dioxide (CO2), which is expelled through the thrusters. SPHERES satellites are powered by AA batteries. The metrology subsystem provides real-time position and attitude information. To simulate ground station-keeping, a laptop will be used to transmit navigational data and formation flying algorithms. Once these data are uploaded, the satellites will perform autonomously and hold the formation until a new command is given. SPHERES is an ongoing demonstration. Each session tests progressively more complex two and three-body maneuvers that include docking (to fixed, moving, tumbling targets), formation flying and searching for lost satellites.

^ back to top


Space Applications
Information learned from this experiment may lead to simpler autonomous docking allowing for servicing, re-supplying, reconfiguring and upgrading of space systems. The algorithms would eliminate the complicated maneuvers that require ground teams to coordinate and execute. Secondly, this formation flight technology would lead to Separated-Spacecraft Interferometers (light from two or more telescopes that combine to provide a high resolution image).The results will support the development of autonomous spacecraft to carry out a variety of tasks in a space environment. Smaller autonomous spacecraft could, with the right coordination and programming, perform tasks too complicated or too expensive for larger spacecraft to execute. Examples of these are satellite clusters, a collection of microsatellites that operate cooperatively to perform the function of a large single satellite.

Earth Applications
The space technologies for formation flight of small satellites could influence Earth-based applications of current satellite technologies including surveillance, mapping, communications and navigation.

^ back to top


Operational Requirements and Protocols
SPHERES has been allocated a total of 28 test sessions. These sessions are nominally 3.5 hours long with approximately eight to twelve tests, each lasting 10-15 minutes. Operations are not permitted in the Russian segment due to concerns with infrared interference. The first 15 test sessions have occurred in Destiny. Future expansion might allow operations in Harmony, Kibo and Columbus.
During the flight sessions there are three phases: programming SPHERES, free-flying operations, and data retrieval. Programming the satellites will involve uploading the algorithms for each specific session to the SPHERES laptop from the ground crew. The laptop will be used to send the algorithms and commands to the satellites and receive data and status reports from the satellites. The data will then be downlinked to the ground crew for analysis. During free-flying operations the satellites will perform various maneuvers with one to three satellites operating simultaneously. Once the test session is complete, the data will be downlinked to the ground, via the Ops LAN, for analysis by the SPHERES team. This analysis will allow new and/or modified tests to be uplinked for use in the next test session. The crew will be responsible for unstowing the equipment, setting up the test area, loading the carbon dioxide tanks and batteries, uploading and running the protocols from the laptop and stowing the equipment at the conclusion of the session.

^ back to top

Decadal Survey Recommendations

Information Pending

^ back to top

Results/More Information

The first few test sessions verified the functionality of many general utilities applicable to a variety of space missions. While continual improvements are being made to general algorithms on SPHERES, specialized algorithms are also being developed and tested to further the areas of autonomous docking and formation flight.

General testing: The first test session, run during Expedition 13, demonstrated operability of all the hardware and the ability to upload programs to the satellites. In this and subsequent test sessions, many modular algorithms such as controllers (calculate desired forces and torques to achieve a certain position and attitude), mixers (convert desired forces and torques into thruster firing times), estimators (calculate the current position and attitude of the satellite) and fault detection and isolation algorithms were developed (Nolet et. al. 2007; Saenz-Otero and Miller 2007).

Many operational and programmatic difficulties were encountered in the initial test sessions (Mohan et. al. 2007). These experiences established desired session frequency of three to four months (as opposed to the original planned frequency of two weeks) to enable full data analysis and algorithm iteration. Real-time audio and video allowed the team to react to problems and implement planned contingencies. New test-plan standards were created to allow the crew to run tests sequentially with a clearly defined success threshold. Feedback from the crew before, during, and after operations resulted in improvements to the procedures and information delivery methods.

During the first five test sessions, an on-board navigation system was developed (Nolet 2007). An extended Kalman filter (also referred to as an estimator) is used to combine data from ultrasound receivers and gyroscopes into an accurate estimate of position and attitude. Two estimators (one using a single beacon and one using all five beacons) were developed and verified. A fault detection scheme was also developed to remove erroneous ultrasound measurements that caused large errors between estimated and actual position and attitude.

In addition to the contributions to docking and formation flight presents the development of closed-loop delta velocity (change in translational/rotational velocity) mixer (Saenz-Otero et. al. 2008). The simple mixers developed in earlier tests were open-loop, meaning they would open and close thrusters in a predetermined fashion that may not accurately produce the desired delta velocity. Instead, the closed-loop delta velocity mixer will use accelerometers and rate gyros in the satellite that can measure the delta velocity in real time to control thrusters. Results from test sessions 10 and 11 show an increase in delta velocity accuracy, but reveal an overshoot in the estimator. An improved estimator was developed in response for test session 12, which incorporated accelerometer and rate gyro data into the calculations.

These general tests produce an overall improvement in SPHERES operations and performance. All subsequent test sessions run smoother with the lessons learned from each test session. In addition, the tools developed enable more complicated tests such as docking and formation flight.

Autonomous docking, inspection and assembly: Docking is a complicated procedure with applications such as on-orbit servicing and autonomous assembly. Nolet et. al. (2007) details the development of the guidance, navigation and control algorithms for docking. This includes estimators, mixers, controllers and fault detection and isolation algorithms. The final integrated tests of these algorithms culminated with the successful demonstration of docking to fixed and tumbling targets, a space first in the case of a tumbling target.

Docking is a high-risk endeavor, which may result in the damage or destruction of both spacecrafts involved if a failure occurs. Fejzic et. al. (2008) explores off-nominal docking scenarios with two different algorithms: a traditional glideslope algorithm and a safe docking algorithm. The glideslope algorithm commands the satellite to decrease its approach velocity linearly with the distance-to-go. The safe docking algorithm calculates a path that guarantees passive collision avoidance, meaning the satellites can be commanded to drift at any point before terminal docking without any chance of collision. Results from test sessions 5 and 6 are presented, showing successful collision avoidance with simulated failures at random times.

Docking may be used in applications such as satellite servicing, refueling and autonomous assembly. In the case of assembly, a tug must dock to a payload then maneuver with the payload. The combined tug-payload system presents challenges such as large changes in mass, inertia and center of mass, thruster configuration and plume impingement. Mohan and Miller (2008) describes reconfigurable mixers that allow smooth transitions in control architectures for various mass configurations. Results from test sessions 6 through 11 are shown, demonstrating the improved performance using the new mixers.

In order to safely operate in close proximity to an object in space, it is important to inspect the target object to determine its position, orientation and three-dimensional geometric map for relative navigation in future stages of the mission. Tweddle et al. (2014) used successfully in test session 38 a Simultaneous Localization and Mapping (SLAM) software algorithm to estimate the static and dynamic states of a free spinning and uncooperative target satellite.

Following the successful addition of new docking and reconfiguration hardware to the SPHERES testbed in early 2016 (Sternberg 2016), additional docking and reconfigurable control algorithms (Jewison et al. 2014) will be tested in 2016 and 2017.  These docking and reconfiguration tests will ultimately enable scientists to assemble large space structures and make autonomous resupply of consumables and upgrades a reality.

Formation flight: Multiple small satellites can mimic the functionality of a single monolithic satellite by distributing tasks to each satellite. One specific application of formation flight is in separated-spacecraft interferometry that can create higher-resolution images than that of a large satellite. Many of the tests apply to formation flight in general while some specifically address the additional challenges of interferometry missions.

One of the first challenges to using multiple spacecraft is formation initialization. Mandy et. al. (2007) shows the results from test session 7 for the first segment of formation initialization: satellites lost in space. A formation must be initialized in cases such as release from the launch vehicle. Without the help of systems such as GPS, satellites must use on-board sensors with a limited field-of-view to locate each other. Results show successful results for two satellites lost in space. Tests are currently being made to demonstrate three satellites lost in space with arbitrary initial conditions.

Once the satellites have found each other, they must begin their formation flight. Because the satellites are likely to join the formation in random order, it is important that each satellite be independently capable of creating a formation or joining an existing one. Saenz-Otero et. al. (2008) and Saenz-Otero et. al. (2009) describe results from test sessions 10 and 11 for random initialization. Tests were only partially successful due to satellite communication issues and unintended initializations. These problems indicate that formation initialization is not a trivial task.

The cyclic pursuit algorithm solves the issue of formation initialization. This algorithm is a decentralized algorithm (there is no leader controlling the movement of each satellite) that tells the satellites to pursue each other. Satellites can be added to the array at any time with guaranteed a convergence to circular, elliptical and spiral formations. Saenz-Otero et. al. (2009) shows results from test session 14 that demonstrates two satellite formations and a seamless transition when a third satellite is added. Further tests are planned for translation of the entire formation as well as satellite rejection in the case of a failure.

For nominal operations of formations in close proximity, it is desirable to have a simple, always-active control law to prevent potential collisions. Saenz-Otero et. al. (2009) describes the results of such a collision avoidance law. This uses a simple technique that maximizes satellites' closest point of approach through thruster firing if two satellites are on a collision course. Successful results are shown for multiple crossings when satellites are commanded to pass through the same point, though one instance of grazing contact was observed.

Two more tests were performed in 2009 to further development of and test the collision avoidance controller, as described in by Katz et. al. (2011). The first test verified two key attributes of the avoidance controller. First, as described in the paragraph above, the algorithm exploits small separations in the projected positions of closest approach and increases them to achieve a safe trajectory. Second, the avoidance controller successfully overcame the nominal controller’s attempt to keep the satellite on a trajectory that could have caused a collision, by disabling the nominal controller for a few seconds at the closest point of approach. Once the danger of a collision past, the nominal controller resumed control of the satellite and continued to guide it to its goal position. The second test was a repeat of the 3-satellite test from the previous session in which the grazing even occurred. The repeated test was set up with new parameters to prevent another occurrence of contact and was a success.

During formation flight, unexpected events such as failures may occur. In such cases, it is important to separate satellites to safe positions. A scatter maneuver was developed and tested in sessions 10 and 11 as described in Saenz-Otero et. al. (2008) and Saenz-Otero et. al. (2009). This maneuver commands all spacecrafts to separate from each other to a safe position that is determined dynamically at the time the command is issued. Final results from test session 11 showed successful scattering to positions that were directly away from all other satellites. Saenz-Otero et. al. (2009) also presents results from test session 14 that successfully use the scatter maneuver to avoid a satellite that has a failed thruster.

In the case of separated-spacecraft interferometry missions, the satellites are required to rotate in formation. The two imaging formations that have been researched thus far are circular and spiral formations. Saenz-Otero et. al. (2009) discusses results from test sessions 7, 12 and 14. A successful circular formation was performed in session 7 with relative position errors within requirements. Spirals were attempted in test sessions 12 and 14. Problems with a new controller caused the failure of tests in test session 12. The tests were rerun with the fixed controller; however the controller still had issues with tracking these spiral trajectories. Future tests are planned to improve tracking spiral trajectories as well as arbitrary trajectories for formation flight.

After measurements have been made of a certain target, the satellite array must be realigned to a different target. Mandy et. al. (2007) outlines the theory for time-optimal, fuel-optimal and fuel-balanced realignment as well as successful results from test sessions 7 and 8. These tests treat attitude control as decoupled from position control. The use of certain sensors, however, can couple not only position and attitude of a single satellite but also multiple satellites. Aoude et. al. (2008) treats an array reconfiguration as an optimization problem with various constraints. Position constraints include avoiding collisions between satellites and other objects. Attitude constraints include keeping sensitive optical sensors pointed away from the sun and relative sensors of the satellites pointed towards each other. Successful results from test sessions 6 and 8 demonstrate the validity of the off-line computation of these path-planning algorithms as all constraints were met throughout the entire reconfiguration maneuver.

The development of these formation flight algorithms in a risk-tolerant environment allows incremental advancement to an otherwise very complex problem. The algorithms developed as well as lessons learned in these test sessions are a large step towards future formation-flying missions.

Tele-operations: Tele-operation of space robots during servicing and maintenance operations provides flexibility and robustness during mission operations when added to existing autonomous systems. Stoll et. al. (2011) describes the results from an investigation testing human operations of the satellites with varying levels of autonomy.

The case study began with a fly around of the inspector satellite around the servicer, which revealed that human operators had difficulty judging the orientation and movement patterns of the satellites in three dimensions.

The Manual Abort test, which evaluated the human perception of a potential collision, showed that the crew had excellent awareness of when collisions were imminent and that a manual abort command could add an additional layer of safety during mission operations. This test also demonstrated that distances were more accurately judged by human operators in three dimensions than was orientation of the satellites.

The Human Navigation test simulated the close approach of an inspector satellite to a target location (such as a particular area of interest during inspection of a spacecraft) using manual commands. The test showed that manual steering of the inspector to another location could be executed precisely and efficiently; however, overall performance was decreased when the inspector was maneuvered indirectly to the goal via a waypoint (in this case, the test volume center). Time and fuel efficiency may be increased by providing the operator with additional reference points with which to aid navigation.

The final tests incorporated a communications delay to simulate the delay in a satellite’s response to ground issued commands. As the length of the delays increased, more time and translation commands were required from the crew to allow the inspector satellite to reach the final goal position. Despite the increase in time and decrease in overall performance, it was demonstrated that manual operations involving a delay in communications can be performed efficiently. Additionally, the inclusion of assisting elements such as autonomous collision avoidance enhanced the operator’s performance by allowing the operator to concentrate on the assigned task rather than worry about collision avoidance.

While increased autonomy is necessary to future space missions, human supervision will continue to be vital. In the future, teleoperations of robotic spacecraft can be used to conduct repairs, perform maintenance, carry out inspections and monitor spacecraft and space operations. Additionally, teleoperations can be used to de-orbit or transfer malfunctioning or defunct spacecraft to the graveyard orbit by ground-based operators.

ASTRO: The SPHERES satellites were used to test a new navigation algorithm called ASTRO (Admissible Subspace Trajectory Optimizer) for real-time path planning and optimization with autonomous free-flying robots.  A series of test runs were performed to validate this technique with progressively more challenging geometries that included moving obstacles.  This research is aimed at developing fully autonomous 3D navigation of robots in complex and dynamic environments to perform various tasks. (Chamitoff 2014).

^ back to top

Results Publications

^ back to top

Ground Based Results Publications

^ back to top

ISS Patents

^ back to top

Related Publications

^ back to top

Related Websites
MIT Space Systems Laboratory
The Register - Space station sphere-sats to collect rocks from Mars

^ back to top


image SPHERES hardware in operation on the KC-135. Flights were in July/August 2002 and February 2003.
+ View Larger Image

image NASA image: ISS018E006422 - Expedition 18 crewmember Michael Fincke as he works with Synchronized Position Hold Engage Reorient Experimental Satellites (SPHERES) in the US Laboratory.
+ View Larger Image

image SPHERES air table testing at Marshall Space Flight Center.
+ View Larger Image

image NASA Image: ISS0008E19135 - Expedition 8 Commander Michael Foale holds the SPHERES Ultrasound Beacon and Beacon Tester while performing functionality checks between the Beacon/Beacon Tester and its proximity to a general luminaire assembly in Unity Node 1.
+ View Larger Image

image NASA Image: ISS008E19136 - Expedition 8 Commander Michael Foale works with the SPHERES Ultrasound Beacon and Beacon Tester while performing functionality checks between the Beacon/Beacon Tester and its proximity to a general luminaire assembly in Unity Node 1.
+ View Larger Image

image Video Screen Shot of Expedition 13 Science Officer, Jeff Williams as he closely monitors the first test flights of the SPHERES satellite in the U.S. Lab Destiny on May 18, 2006.
+ View Larger Image

image NASA Image: ISS013E65815 - Expedition 13 NASA Science Officer, Jeff Williams, carrying out test runs with two satellites on August 12, 2006. SPHERES began with a single satellite floating inside the cabin and going through iterative tests to "learn" how to maneuver in the cabin. STS-121 carried the second satellite to ISS for testing.
+ View Larger Image

image NASA Image: ISS013E68304 - Astronaut Jeffrey N. Williams, Expedition 13 NASA space station science officer and flight engineer, does a check of the Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) satellites in the Destiny laboratory of the International Space Station.
+ View Larger Image

image NASA Image: ISS014E08025 - View of the Synchronized Position Hold,Engage,Reorient,Experimental Satellites (SPHERES) floating in the Destiny laboratory module as seen by the Expedition 14 crew. Flight Engineer Thomas Reiter is visible in the background.
+ View Larger Image

image NASA Image: ISS014E17232 - Astronaut Michael E. Lopez-Alegria, Expedition 14 commander and NASA space station science officer, does a check of the Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) Beacon / Beacon Tester in the Destiny laboratory of the International Space Station. SPHERES demonstrates the basics of formation flight and autonomous docking, using beacons as reference for the satellites, to fly formation with or dock to the beacon.
+ View Larger Image

image NASA Image: ISS020E018324 - NASA astronaut Michael Barratt (left) and Japan Aerospace Exploration Agency (JAXA) astronaut Koichi Wakata, both Expedition 20 flight engineers, perform a check of the Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) Beacon / Beacon Tester in the Destiny laboratory of the International Space Station.
+ View Larger Image

image NASA Image: ISS014E17874 - Three satellites fly in formation as part of the Synchronized Position Hold, Engage, Reorient, Experimental Satellites (SPHERES) investigation. This image was taken during Expedition 14 in the Destiny laboratory module.
+ View Larger Image