NASA JSC Active Response Gravity Offload System

ARGOS is the gravity offload facility at NASA JSC: an overhead bridge crane carrying a motor-driven cable that holds a constant upward force on a test article while following it through the workspace, so the article moves as though it weighed a lunar, Martian or microgravity fraction of its Earth weight [1][4]. Three translational axes are actively driven, which is the difference between it and the earlier Pogo rig described on the reduced gravity testing page, and the reason it can be used for walking rather than only for standing tasks [2].

It offloads at one point. Everything the following sections say about fidelity follows from that: the cable knows the total weight it is holding, and nothing about how the mass is distributed across the limbs beneath it [3].

Every number in the capability table is the operator’s stated specification. Nothing published reports a measured offload force accuracy, a bandwidth, a residual horizontal force or a cable angle error against an independent reference [1][4].

Laboratory data

Parameter	Value
Operator	NASA JSC [4]
Location	Houston, Texas, United States [4]
Commissioned	Third iteration of the system as described in 2017; earlier dates not published [1]
Type	Active cable gravity offload rig, human rated and two fault tolerant [1]
Floor area	Bridge crane structure 41 x 24 x 25 ft (12.5 x 7.3 x 7.6 m) [4]
Capabilities	Offload envelope, gimbals
Simulant or terrain	None installed. Terrain is brought in by the test [1]
Instrumentation	Displacement and force sensing on X, Y and Z; AIBEL motion tracking [4]
Ground truth	Not published. No independent verification of delivered offload is reported [1][4]
Fidelity limits	Single attachment point; limbs stay at 1 g; three rotations passive [2][3]
Access	Test support offered through the JSC analogs and mockups program; no lead time published [4]
Cited by	Robonaut 2 [5]

Capabilities

Offload envelope

Parameter	Value
Working volume	Over 340 m3: 11.0 m Y, 5.8 m X, 5.5 m working height [1]
Test article limits	340 kg (750 lb) payload [1][4]
Vacuum	None. Ambient pressure in a high bay [1]
Temperature	Not controlled. Breathing air and suit cooling water are supplied [4]
Illumination	Not published. No solar simulator
Slope	Not applicable. Terrain is whatever the test brings in
Gravity offload	Vertical, at one point, set in software from 1 g to microgravity [1][4]
Instrumentation	Redundant sensors, real-time safety model, independent overspeed detection [1]

Speed and acceleration trade against payload mass: 3.3 m/s and 24.4 m/s2 at 140 kg, falling to 1.4 m/s and 10.2 m/s2 at 340 kg [1]. A separate horizontal-system maximum of 2.0 m/s and 6.1 m/s2 is quoted alongside those figures without being reconciled with them [1]. The operator’s own catalog gives the limits in the form a test director needs, as tracking rates by configuration: 4 ft/s vertical and 6.5 ft/s horizontal with a suited subject, 11 ft/s vertical and 6.5 ft/s horizontal unsuited [4]. Motion faster than that is outside the offload envelope, which is a constraint on the task, not on the rig.

Vertical offload load sensitivity is quoted as below 0.0005 kg, and horizontal tracking sensitivity as under 0.1 degree of cable angle, both without a bandwidth, a condition or a measurement method [1]. Safety faults are detected in under 20 ms with two fault tolerance, and a lifting cable device limits the shock load into the payload during a high-speed stop [1].

The two published descriptions disagree on the workspace. The 2017 description gives 11.0 by 5.8 m at 5.5 m height, over 340 m3 [1]; the operator’s catalog gives 13 by 30 by 15 ft, about 4.0 by 9.1 by 4.6 m or 167 m3, inside the 41 by 24 by 25 ft crane structure [4]. Neither states which convention it uses, and the difference is a factor of two in usable volume, so a test that needs the full run length should confirm it with the facility rather than from either number here.

Gimbals

Parameter	Value
Working volume	Not applicable. The gimbal is the payload interface
Test article limits	Suited, unsuited and unmanned interfaces available [4]
Actively controlled axes	Three translations. The three rotations are passive [2]
Instrumentation	Center of gravity board with four-corner force transducers, for pivot setup [3]

Roll is limited to about 30 degrees by the parallelogram linkage and pitch by interference with the subject [2]. Gimbal mass is a facility variable with a measurable effect: replacing the 105 lb all-metal claw gimbal with a 53 lb gimbal adapted from EMU and MkIII suit hardware measurably improved subject-rated simulation quality, the lower mass and inertia being the stated cause [2].

What it does not reproduce

The limbs. The gimbal attaches at a single point on the portable life support system mockup, so arms and legs move against full 1 g. It is worst in prone postures, where the entire lower torso lies outside the offloading cable region [3]. Any result about limb effort in a suited task at ARGOS is a 1 g limb result. This is a fidelity limit stated by the operators, not a measurement of the error it causes, and no published source quantifies it.

Gait timing. Suited walking at 1/6 g offload gave a 50/50 swing and stance split, the same as 1 g unsuited ground walking, but the mid-swing point occurred 10 percent earlier and the cause was not established [2]. Different subjects were used in the two environments, so the comparison is not paired.

A pivot point that does not have to be tuned. Across 68 configurations and 14 xEMU subjects of 175.1 +/- 8.4 cm stature in lunar offload, the functional zone for the gimbal pivot is 2.5 to 6.4 cm above and 1.3 to 2.5 cm aft of the system center of gravity [3]. Suited subjects notice fore-aft shifts as small as 0.6 cm, while up-down shifts of 2.5 cm produce minimal performance difference [3]. A simulation run without that tuning is measuring the rig. Setup uses a center of gravity board with four-corner force transducers and the hard upper torso hatch angle: standing neutral came out at 58 to 62 degrees from vertical, standing tip-over at 47 to 52 degrees, one-knee kneeling tip-over at 15 to 27 degrees, with a standing sway range of 0.5 to 8.5 in and a static center of pressure offset of 3 to 3.5 in [3].

Soil. The rig offloads the test article. The ground beneath it stays at 1 g, so a mobility or excavation result taken here inherits terrain stronger than the destination’s [1]. The reasoning is on the reduced gravity testing page.

Duration is the thing it does reproduce. Against a parabolic aircraft, which gives the highest offload fidelity of the analogs but only about 30 s per parabola in a constrained cabin volume, ARGOS trades fidelity for unlimited run time [2]. Against Pogo, which drove only two axes and carried significant overhead inertia from the lifting mechanism, it trades mechanism complexity for a third driven axis [2].

Campaigns run there

Robonaut 2 with climbing legs, 2014. A nearly 300 kg two-legged Robonaut 2 configuration placed its end effector to under 1 cm of error over a reach of more than 1 m, with both legs stiff at a natural frequency of 20 and a damping ratio of 1.2, base leg joint torque limited to 100 and reaching leg to 20 in units the source does not state, and the embedded impedance loop integrator enabled [5]. The result carries the facility’s own limit: only the base leg was effectively offloaded, so the reaching leg worked against 1 g [5].

xEMU pivot point characterization, reported 2023 [3]. The 68-configuration study above is a campaign about the facility rather than about a test article, and it is an unusual thing for an offload rig to publish. One dedicated run tested 10 pivot settings on a single experienced subject in a day [3].

Suited lunar gait measurement, reported 2020. Inertial measurement units mounted on the exterior of a Mark III suit recorded the swing and stance split at 1/6 g offload [2].