Skip to content

The JEMRMS main arm, upper right, holding the Interorbital Communication System payload after unberthing it from the Kibo Exposed Facility during STS-127 on 23 July 2009. The two CFRP booms and the wrist joint cluster stand against the Kibo pressurized module; the thicker boom entering from the lower right is Canadarm2. NASA. Public domain (NASA / US government work).

The JEM Remote Manipulator System is the robotic system of Kibo, the Japanese Experiment Module of the International Space Station. It comprises two arms and a console inside the pressurized module [1]. The main arm handles payloads on the Exposed Facility, the small fine arm, and Kibo elements such as the Experiment Logistics Module Exposed Section. The small fine arm attaches to the tip of the main arm and performs dexterous work such as orbital replacement unit exchange [1], [2]. It was developed by the National Space Development Agency of Japan with Toshiba as prime contractor for the arm hardware [1], [3], and is operated by JAXA [7]. Contemporary Japanese reviews list it as the agency’s principal space robotics project of the 1990s [4].

Both arms have six degrees of freedom in an anthropomorphic layout of two shoulder joints, one elbow joint and three wrist joints [1]. The main arm carries three booms, two long booms of carbon fiber reinforced plastic chosen for low thermal expansion, high stiffness and low mass, and one short aluminum boom. The small fine arm carries two aluminum booms and is 2.2 m long and 190 kg [7]. Every joint on both arms has a brake that is active without electrical power [1].

ParameterMain armSmall fine armSource
Degrees of freedom66[1], [7]
Joint layout2 shoulder, 1 elbow, 3 wrist2 shoulder, 1 elbow, 3 wrist[1]
Length10 m2.2 m[7]
Mass780 kg190 kg[2], [7]
Rated payload7,000 kg300 kg, or 80 kg under force moment accommodation[2], [7]
Positioning accuracy, translation+/-50 mm+/-10 mm[1], [7]
Positioning accuracy, rotation+/-1 deg+/-1 deg[1], [7]
Maximum tip forceover 30 Nover 30 N[1], [7]
Tip translation rate60 mm/s under 600 kg, 30 mm/s to 3,000 kg, 20 mm/s to 7,000 kg50 mm/s under 80 kg, 25 mm/s to 300 kg[7]
Maximum stopping distance, translation300 mm75 mm[1]
Maximum stopping distance, rotation5 deg5 deg[1]
Joint output torqueabout 400 N mabout 41 N m joints 1 to 3, 18 N m joints 4 to 6[2]
Joint reduction ratio1400:1130:1 joints 1 to 3, 120:1 joints 4 to 6[2]
Joint brake torque150 to 270 N m40 to 120 N m[2]
Joint rotational stiffnessabout 3e5 N m/rad0.8e4 to 1.6e4 N m/rad[2]
Joint bending stiffnessabout 1e6 N m/radnot published[2]
Joint backlash0.043 to 0.056 deg0.03 to 0.06 deg[2]
Housing alignment uncertaintywithin 0.012 deg0.01 to 0.03 deg[2]
Design lifeover 10 yearsnot published[7]

The 1994 design specification gives lower rates than the flight figures above, with main arm translation of 10 mm/s under a 7,000 kg payload rising to 60 mm/s under 600 kg, and rotation of 0.5 deg/s under 7,000 kg rising to 2.5 deg/s under 600 kg [1]. The small fine arm design figures are 10 mm/s and 0.5 deg/s under 300 kg, and 30 mm/s and 1.0 deg/s under 130 kg.

Natural frequencies of the main arm were computed by finite element model in a near extended reference configuration with the elbow pitch joint at 150 degrees to avoid the singularity [1]. The storage configuration was chosen both to lower heater power and to raise the natural frequency above that of the reference configuration, the second consideration mattering because the arm is a 10 m, 780 kg structure [7], [2].

Configuration1st mode2nd mode
On orbit, no payload0.32 Hz0.34 Hz
On orbit, 500 kg payload0.19 Hz0.21 Hz
On orbit, 7,000 kg payload0.046 Hz0.05 Hz
On orbit, storage configuration0.65 Hz1.01 Hz
Launch configuration14.2 Hz14.9 Hz

On-orbit natural frequency is set mainly by the torsional stiffness of the joints, which is in turn set mainly by the torsional stiffness of the speed reducers [1]. Boom stiffness is much larger than joint stiffness on both arms, so the joints dominate arm compliance [2].

EventDate
Three payloads assembled onto the Exposed Facility by astronaut teleoperation of the arm2009-07
Two further payloads assembled the same way2009-09
Exposed Pallet assembly and removal, first occasion2009-09
Exposed Pallet assembly and removal, second occasion2011-02

Source: [6].

Through February 2011 the arm and the Exposed Facility Unit had assembled five payloads and had berthed and removed an Exposed Pallet twice, all under astronaut teleoperation [6]. The pair could not be verified together on the ground with flight hardware, because the arm cannot execute its three-dimensional motion under gravity: a hardware simulator reproducing the arm tip positioning performance supplied the basic behavior data for the coordinated operation, and a software simulator of arm motion and contact covered the off-nominal cases that cannot be reproduced on the ground. Before each payload was assembled the arm’s vision system was squared onto the vision marker of the target Exposed Facility Unit and imaged, and ground image measurement of the unit’s position found corrections from the design values larger than expected, up to 10 cm [6]. The Kibo standard payload against which the main arm was sized is 500 kg and 1 m by 1 m by 1.9 m, berthed to one of twelve Exposed Facility Units [2].

The main arm consists of three booms, six joints, a base, an end effector and two vision equipment assemblies [1]. The base is titanium and carries a curvic coupling so that an EVA crew member can separate the arm from the pressurized module for maintenance. The end effector is similar to the standard end effector of the Shuttle Remote Manipulator System, and grapples the same grapple fixture interface used elsewhere on the station [1], [2]. JAXA describes it as a grapple fixture interface for Exposed Facility payloads, and the arm is rated for a design life of over ten years [7].

The small fine arm consists of two aluminum booms, six joints, an electronics unit, an end effector, a force and moment sensor and a TV camera [1]. Its end effector is a Kibo-unique device called the Tool, which grasps a Tool Fixture on the payload and supplies torque to a bolt through it [1], [2]. The prototype built for the JEM Flight Demonstration used a three-finger tool with an integrated torque driver for loosening and fastening the bolts that structurally attach an orbital replacement unit [3]. A grapple fixture on the base of the small fine arm is what the main arm holds when the small arm is in use [2].

Kibo itself comprises the pressurized module, the Exposed Facility, the Experiment Logistics Module pressurized and exposed sections, and the manipulator system [1], [8]. In the launch configuration the main arm is stowed on the aft end plate of the pressurized module by three hold and release mechanisms [1]. Pressurization of the module deforms it and forces a relative displacement on the stowed arm, which would induce large loads, so one of the three mechanisms carries a load relief device in the boom axial direction.

Each of the six main arm joints is a revolute unit delivering about 400 N m of output torque from a brushless DC motor through solid lubricated gearing at 1400:1 [2]. The joint carries a magnetic resolver of 0.002 degree resolution on the motor axis and an optical absolute encoder between the inner and outer housings of 0.006 degree resolution and 0.011 degree accuracy. Main bearings are solid lubricated under constant preload, and an off-activated ceramic brake sits on the motor axis. Each joint contains a joint electronics unit that closes the angle and angular velocity loops locally [1].

The small fine arm has two sets of three identical joints, joints 1 to 3 delivering about 41 N m and joints 4 to 6 about 18 N m, each through a solid lubricated Harmonic Drive at 130:1 and 120:1 respectively [2]. Position feedback is by magnetic encoder of 0.006 degree resolution. The rated handling mass of that arm is 300 kg, falling to 80 kg when force moment accommodation is in use [7].

Back-drive torque matters for the main arm because payload mating and demating is performed by letting external force move the arm tip in a limp mode, and the resisting torque is dominated by gear friction [2]. It was measured at 10.4 to 20.3 N m in thermal vacuum and 17.5 to 39.0 N m in nitrogen gas at ambient pressure. On the small fine arm the same quantity is less important because contact operations are actively limped through the force and torque sensor, and was measured within about 10 N m for joints 1 to 3 and 8 N m for joints 4 to 6, excluding up to about 10 N m of cable loss torque in orbital conditions [2].

The small fine arm carries an aluminum alloy six-axis force and torque sensor at its tip, which is what enables compliance control and active limp control [2].

ChannelRangeAccuracyResolution
Forceto 10 Nabout 1 N0.17 N
Forceto 60 Nabout 2 N0.17 N
Momentto 1 N mabout 0.1 N m0.015 N m
Momentto 10 N mabout 0.3 N m0.015 N m

Values from [2], measured after overload force and moment tests, against full ranges of 60 N and 9 N m. Rated tip force for both arms is over 30 N [7].

Both arms are controlled by dual 32-bit microprocessors located in the pressurized module, communicating over dual MIL-STD-1553B networks [2]. The control algorithm is classical, using phase lead and lag filters, and the controller is closed on the motor axis rather than the output axis of the speed reducer so that joint backlash is kept outside the closed loop [1], [2]. Eight sets of control parameters are stored in each joint electronics unit and selected by the arm control unit according to the payload mass properties and the arm configuration [1]. The small fine arm additionally runs five 32-bit microprocessors in an electronics box at its base, which close the angular velocity loops of its six joints [2].

Arm performance was predicted before flight by a non-real-time computer simulator running the controller against the joint characteristics, and the mathematical model was updated from modal survey and two-dimensional flat floor function tests [1].

The main arm is operated primarily in preprogrammed control, with manual control through the rotational and translational hand controllers as the alternative [1]; both controllers are part of the console in the pressurized module [7]. For the small fine arm manual control is primary [1]. In manual mode the crew member works from joysticks with overlay templates superimposed on the display [2].

Preprogrammed motion relies on vision to correct for misalignment. Kibo assembly tolerances and thermal distortion move the target relative to the arm base, so the position and orientation of a target are measured by the wrist vision equipment and the management data processor and fed into the preprogrammed path [1]. Visual sensing uses a three-dimensional four-point target at the worksite [2]. The crew member completes the calibration by adjusting four cursors on the screen until they overlap the four target points. Visual calibration carries many error sources and is the dominant contributor to total positioning error.

Berthing a payload to an Exposed Facility Unit follows an L-shaped approach path rather than a straight one, to avoid jamming with the capture mechanism [2]. The wrist camera pan and tilt capability exists so that the arm can be positioned at the coarse visual calibration point and then at the fine positioning point at the Payload Interface Unit even when a large payload blocks the view ahead of the arm.

Positioning error of the main arm was analyzed as a total of under 44.7 mm in translation and 0.46 degrees in rotation, built from joint backlash at under 20.9 mm, housing alignment at under 16.4 mm, assembly at under 16.3 mm, control error at under 5.3 mm and thermal at under 2.2 mm [2]. Ground measurement by three-dimensional measuring equipment gave 22.9 mm in x, 54.0 mm in z and 0.71 degrees. Eliminating ground friction gives an on-orbit estimate of 14.7 mm, 29.2 mm and 0.370 degrees on output-axis encoder data [2]. The specification the arm is held to is +/-50 mm in translation and +/-1 degree in rotation [1], [7].

QuantityMain armSmall fine arm
Positioning error, own coordinate systemunder 44.7 mm, under 0.46 degunder 2.62 mm, under 0.69 deg
Positioning error mounted on a servo-locked main armunder 8.28 mm
Deviation from nominal path, along path50.9 mm test, 52.1 mm analysis26.2 mm analysis
Deviation from nominal path, radial16.4 mm test, 14.3 mm analysis6.3 mm analysis
Emergency stop, 600 kg at 60 mm/s117.0 mm test, 129.6 mm analysis, 0.39 deg test
Emergency stop, 3,000 kg at 30 mm/s89.5 mm test, 121.6 mm analysis, 0.51 deg test

Source: [2].

System performance was compared between worst-case analysis and system test data taken on an air bearing floor in two-dimensional arm movement, with the small fine arm always evaluated attached to a servo-locked main arm because that is how it is used [2]. A two-dimensional flat floor test was part of the engineering model program from the outset [1], and the small fine arm was designed from the start to be carried on the main arm tip [1], [4]. Measured emergency stop distances stay inside the 300 mm and 5 degree limits set in the design specification [1].

Berthing conditions were verified against the design specification limits.

Berthing quantityLimitEvaluated
Forward approach speed30 mm/s10 mm/s
Radial approach speed30 mm/s
Rotational rate0.5 deg/s
Effective mass at the main arm wrist400 kg320 kg
Attitude within the approach envelope3 degrees1.7 degrees
Capture wobble1.6 degreesunder 1.4 degrees
Axial capture angle1.2 degreesunder 0.6 degrees
Back-drive force, axial110 N43 N
Back-drive force, radial110 N72 N
Back-drive torque, rotational220 N m179 N m
Back-drive torque, wobble220 N m196 N m

Source: [2].

Berthing is the task the arm exists to perform, since the Exposed Facility is loaded and unloaded by the arm rather than by crew [1], [7], and a failed capture would produce debris or damage station structure [2].

The main arm carries two vision equipment assemblies, each a TV camera on a pan and tilt unit; only the wrist assembly has an arm light [1]. The wrist camera has pan, tilt, zoom, iris and focus control and provides the six-degree-of-freedom measurement used for automatic manipulation [2]. The small fine arm carries a fixed-focus camera on its wrist. Camera and light assemblies on the main arm booms are listed by JAXA as part of the delivered system [7].

Both arms are operated by a crew member at the JEMRMS console in the pressurized module [1], [7]. The console comprises two TV monitors, passive rotational and translational hand controllers, hand controller electronics, a remote interface panel, a laptop workstation, a task light, a management data processor, a mass storage unit, an arm control unit, a power distribution box, a rack essential package, a fire detection and suppression panel, hold and release mechanism electronics, and a standard double rack [1]. NASA supplies a split screen capability that lets the operator view three video feeds simultaneously during robotic operations.

A six-degree-of-freedom orthogonal force-reflecting hand controller was developed as a prototype common master for both arms, on the argument that an orthogonal master simplifies the coordinate transformation to any slave configuration and that a single unit therefore suffices for two articulated arms of different geometry [5]. The prototype measured about 350 by 350 by 350 mm with a 150 mm cubic motion envelope and weighed 15 kg including mechanical dead weight compensation. It was tested both as a rate-control joystick and as a force-reflecting master against a six-degree-of-freedom experimental manipulator, the intent being that one master serve both the main arm and the small fine arm from the small operating compartment available. The flight console uses separate passive rotational and translational hand controllers [1].

JAXA lists the console contents as a management data processor, a laptop computer, rotational and translational hand controllers, TV monitors and hold and release electronics [7]. The console layout was itself revised during development. Termination of the multi-purpose application console in the space station redesign forced a redesign of the JEMRMS console, with its concept design review held at the end of October 1994 [1].

Engineering model testing of the main arm was planned from summer 1995 to summer 1996 and covered function testing including a two-dimensional flat floor test, modal survey in both orbit and launch configurations, static load test, random vibration, electromagnetic compatibility and thermal balance [1]. Static load and random vibration were qualification tests. The end-to-end system test with the main arm, the console and the small fine arm was limited to the extensive function test including the two-dimensional flat floor test of the main arm carrying the small fine arm. Qualification results were later published as evidence that the design met its performance requirements [2].

The JEM Flight Demonstration was a separate program to fly a prototype small fine arm in the Space Shuttle payload bay on a multi-purpose experiment support structure, operated by a crew member from a workstation on the aft flight deck, with an orbital replacement unit and a task panel as robotic targets and two camera and light assemblies for visual feedback [3]. Launch was scheduled for 1997. Its preliminary design review completed in December 1992, and phase 0 and phase 1 safety reviews were conducted against Shuttle payload safety requirements [3]. It never flew.

The program produced a large space manipulator built around solid lubricated gearing and bearings characterized for back-drive torque in thermal vacuum, which is the parameter that governs whether a payload can be mated by letting the arm go limp [2]; a tip-mounted six-axis force and torque sensor accurate to about 1 N and 0.1 N m that supports compliance control and force moment accommodation on a dexterous sub-arm; and the pairing of a large arm with a small dexterous arm carried on its own tip and stowed separately on an exposed facility [1]. JEMRMS is identified in contemporary reviews as one of the three Japanese space robotics projects of its era alongside the JEM Flight Demonstration and ETS-VII [4].

References

  1. Matsueda, T., Kuwao, F., Motohashi, S. and Okamura, R. (1994). Development of Japanese Experiment Module Remote Manipulator System. Source
    BibTeX
    @inproceedings{matsueda1994development,
      author = {Matsueda, Tatsuo and Kuwao, Fumihiro and Motohashi, Shoichi and Okamura, Ryo},
      title = {Development of Japanese Experiment Module Remote Manipulator System},
      booktitle = {Third International Symposium on Artificial Intelligence, Robotics and Automation for Space (i-SAIRAS)},
      pages = {183--186},
      year = {1994},
      url = {https://ntrs.nasa.gov/citations/19950017288}
    }
  2. Wakabayashi, Y., Morimoto, H., Satoh, N., Hayashi, M., Aiko, Y. and Suzuki, M. (2002). Performance of Japanese Robotic Arms of the International Space Station. Source
    BibTeX
    @inproceedings{wakabayashi2002performance,
      author = {Wakabayashi, Y. and Morimoto, H. and Satoh, N. and Hayashi, M. and Aiko, Y. and Suzuki, M.},
      title = {Performance of Japanese Robotic Arms of the International Space Station},
      booktitle = {15th IFAC Triennial World Congress, Barcelona},
      year = {2002},
      doi = {10.3182/20020721-6-es-1901.00925},
      volume = {35},
      pages = {115-120}
    }
  3. Nagatomo, M., Tanaka, M., Nakamura, K. and Tsuda, S. (1994). Space Robotic Experiment in JEM Flight Demonstration. Source
    BibTeX
    @inproceedings{nagatomo1994space,
      author = {Nagatomo, Masanori and Tanaka, Masaki and Nakamura, Kazuyuki and Tsuda, Shinichi},
      title = {Space Robotic Experiment in JEM Flight Demonstration},
      booktitle = {Third International Symposium on Artificial Intelligence, Robotics and Automation for Space (i-SAIRAS)},
      pages = {149--152},
      year = {1994},
      url = {https://ntrs.nasa.gov/citations/19950017281}
    }
  4. Ohkami, Y., Nakatani, I., Wakabayashi, Y. and Iwata, T. (1994). A Perspective on Space Robotics in Japan. Source
    BibTeX
    @inproceedings{ohkami1994perspective,
      author = {Ohkami, Yoshiaki and Nakatani, Ichiro and Wakabayashi, Yasufumi and Iwata, Tsutomu},
      title = {A Perspective on Space Robotics in Japan},
      booktitle = {Third International Symposium on Artificial Intelligence, Robotics and Automation for Space (i-SAIRAS)},
      year = {1994},
      url = {https://ntrs.nasa.gov/citations/19950017275}
    }
  5. Matsuhira, N., Iikura, S., Asakura, M. and Shinomiya, Y. (1989). Development of a Multipurpose Hand Controller for JEMRMS. Source
    BibTeX
    @inproceedings{matsuhira1989development,
      author = {Matsuhira, Nobuto and Iikura, Shoichi and Asakura, Makoto and Shinomiya, Yasuo},
      title = {Development of a Multipurpose Hand Controller for JEMRMS},
      booktitle = {NASA Conference Publication},
      pages = {105--116},
      year = {1989},
      url = {https://ntrs.nasa.gov/citations/19900012771}
    }
  6. Ueno, H., Doi, S., Wakabayashi, Y., Kuwao, F., Takegai, T., Yoshie, Y., Takada, S., Hattori, H., Fukui, N. and Watanabe, J. (2012). On-orbit assembly technology by Kibo robot arm and equipment exchenge unit. Journal of the Robotics Society of Japan, 1. Source
    BibTeX
    @article{ueno2011japanese,
      author = {Ueno, Hiroshi and Doi, Shinobu and Wakabayashi, Yasushi and Kuwao, Fumihiro and Takegai, Tomoki and Yoshie, Yuki and Takada, Shoji and Hattori, Hiroaki and Fukui, Norio and Watanabe, Junichiro},
      title = {On-orbit assembly technology by Kibo robot arm and equipment exchenge unit},
      journal = {Journal of the Robotics Society of Japan},
      volume = {30},
      number = {1},
      pages = {49--50},
      year = {2012},
      doi = {10.7210/jrsj.30.49},
      url = {https://www.jstage.jst.go.jp/article/jrsj/30/1/30_1_49/_pdf}
    }
  7. (2026). JAXA: Remote Manipulator System, About Kibo. iss.jaxa.jp/en/kibo/about/kibo/rms (accessed 2026-09-02) archived copy
    BibTeX
    @misc{jaxaremote,
      title = {JAXA: Remote Manipulator System, About Kibo},
      howpublished = {\url{https://iss.jaxa.jp/en/kibo/about/kibo/rms/}},
      organization = {iss.jaxa.jp},
      year = {2026},
      urldate = {2026-09-02}
    }
  8. (2026). JAXA: Major Component, About Kibo. iss.jaxa.jp/en/kibo/about/kibo (accessed 2026-09-02) archived copy
    BibTeX
    @misc{jaxamajor,
      title = {JAXA: Major Component, About Kibo},
      howpublished = {\url{https://iss.jaxa.jp/en/kibo/about/kibo/}},
      organization = {iss.jaxa.jp},
      year = {2026},
      urldate = {2026-09-02}
    }

Further reading

  • NASA. (2020). Cross-Program Design Specification for Natural Environments (DSNE), Revision G. NASA Marshall Space Flight Center. Source
  • Lear, D. M., Hoffman, K. D., Hyde, J. L. and Collins, C. M. (2019). Bumper: A Tool for Analyzing Spacecraft Micrometeoroid and Orbital Debris Risk. NASA. Source
  • O'Neill, P. M., Golge, S. and Slaba, T. C. (2014). Implementing the Badhwar-O'Neill Galactic Cosmic Ray Model for Spacecraft Analysis. NASA. Source