Source-linked AI summary
A Deployable Four-Finger Payload for Teleoperated Free-Flying Manipulation with Astrobee
William Su, Jordan Kam, Yunosuke Nakamura, Yixiao Wang, Jianshu Zhou, Masayoshi Tomizuka
TL;DR
Astrobee’s standard perching arm is limited in dexterous free-flight tasks, while habitat logistics may be needed during periods with limited or no crew. The paper pairs a deployable four-finger payload with bimanual VR control and reports ten trials in an ISS-like simulation.
Problem
Astrobee’s standard perching arm is limited in the range of dexterous tasks it can perform, even in free flight, while habitats need recurring logistics during periods with limited or no crew.
Source: §I · p. 1Method
The paper proposes a deployable four-finger payload with 17 manipulation coordinates and a VR interface that maps hand articulation, wrist separation, and common wrist motion to the fingers, rail, and Astrobee.
Source: §I · p. 1, §II · p. 2, A. Teleoperation Pipeline · p. 2Results
All ten trials established continuous confirmed opposing CTB contact and executed the intended maneuver; median teleoperation duration was 77.5 s and median RMS deviation was 35.6 mm.
Source: §III · p. 3Takeaways & Limitations
ISS-like simulation showed that the pipeline can be used to teleoperate Astrobee in transporting cargo.
Source: §IV · p. 3The integrated system was evaluated in simulation; hardware validation was still in preparation and was not reported.
Source: §II · p. 2
Abstract
from arXiv · showhide
This article presents a bimanual teleoperation pipeline and conceptual design of a deployable four-finger payload for intra-vehicular free-flyers. Future habitats in low-Earth orbit (LEO) will require systems to perform mundane tasks like cargo handling and maintenance during crewed and uncrewed periods. The gripper payload provides 17 manipulation degrees-of-freedom (DoF) through four independently actuated fingers on a linear rail system. To control it, a virtual reality (VR) device interface maps the human ground operator's hand motions to the finger pairs, their separation to the rail, and common wrist motion to Astrobee translation. We present the preliminary results of teleoperating Astrobee in a custom zero-gravity MuJoCo-based International Space Station (ISS) simulator through ten repeated trials of transporting a rigid ISS Cargo Transfer Bag (CTB). We measure task success, continuous contact retention, completion time, and cargo motion.
I. INTRODUCTION
Habitats may need recurring logistics and maintenance during periods with limited or no crew, while Astrobee’s standard perching arm is limited in dexterous free-flight tasks. The paper proposes a deployable four-finger payload and teleoperation pipeline to extend Astrobee’s manipulation capabilities.
I. INTRODUCTION
- Near-term commercial low-Earth-orbit destinations will need recurring resupply, maintenance, and logistics, including during periods with limited or no onboard crew.Free-flying robots could reduce crew time spent on these routine tasks.Source: §I · p. 1
- Prior concepts include gecko-inspired adhesives and hybrid rigid-soft pneumatic grippers; dual-arm systems add reach but increase stowed volume and coordination.For objects Astrobee can approach directly, vehicle motion can provide gross positioning without extended arm reach.Source: §I · p. 1
- The contributions are a deployable four-finger rail gripper, concurrent Quest-based control, and evaluation with rigid-bodied ISS Cargo Transfer Bags; Fig. 1 shows one CTB in simulation.The interface controls the fingers, pair-spacing rail, and robot motion concurrently.Source: §I · p. 1, Figure 1 · p. 1
II. METHODOLOGY
The payload combines four multi-DoF fingers with a spacing rail and a vertical stage that stows the mechanism in Astrobee’s lower payload bays. The VR pipeline concurrently controls finger pairs, pair spacing, and Astrobee motion; the integrated system is evaluated in simulation.
II. METHODOLOGY
- Four fingers provide 16 manipulation DoF, and a linear rail adds one pair-spacing DoF; a vertical stage stows the mechanism in Astrobee’s two lower 1-U payload bays and deploys it below the vehicle.Fingers A and C are fixed, while B and D share the rail carriage.Source: §II · pp. 1–2
- The VR interface maps calibrated hand features to the two finger pairs, inter-wrist distance to the spacing rail, and common wrist translation to Astrobee motion.The commanded manipulation configuration contains the four finger configurations and rail displacement; deployment is separate from those 17 coordinates.Source: Figure 2 · p. 2, §II · p. 2
- The integrated system is evaluated in simulation; hardware validation was in preparation and is not reported.The study therefore reports simulation results rather than hardware validation.Source: §II · p. 2
A. Teleoperation Pipeline
The VR interface maps hand geometry to finger-pair targets, wrist separation to rail spacing, and coordinated wrist motion to Astrobee motion. An optimization-based retargeter converts finger-pair targets into joint commands.
A. Teleoperation Pipeline
- Tracked thumb–index geometry and wrist motion define finger-pair targets, while an optimization-based step converts those targets into commands for 16 finger joints.Each hand controls an eight-DoF finger pair.Source: A. Teleoperation Pipeline · p. 2
- Fingertip separation, pair direction and center, and distal-link directions form the geometric features mapped from each hand to robot-space targets.Pinch-to-open calibration also supplies a posture reference consistent with the commanded opening.Source: A. Teleoperation Pipeline · p. 2
- The retargeting objective tracks span, pair direction and center, and distal-link directions, while posture and temporal terms resolve redundancy and suppress abrupt joint-coordinate changes.The geometric scales and weights are fixed controller settings for all evaluation trials.Source: A. Teleoperation Pipeline · p. 2
- Inter-wrist distance directly commands rail spacing, while coordinated wrist translation steers Astrobee; returning both wrists to neutral commands smooth braking.The left wrist sets travel direction and displacement scales requested speed up to 0.20 m/s.Source: A. Teleoperation Pipeline · p. 3
B. ISS Simulation
The evaluation uses a custom zero-gravity ISS simulation in MuJoCo, with Astrobee and the cargo bag modeled as free rigid bodies. It records teleoperation and motion data and measures cargo movement relative to the vehicle during confirmed contact.
B. ISS Simulation
- The MuJoCo simulation integrates physics at 1 kHz beneath a 50-Hz teleoperation loop and models Astrobee and the CTB as free rigid bodies.The task moves cargo from the Kibo Module to a connected destination module through acquisition, translation, and a turn.Source: B. ISS Simulation · p. 3
- At each 50-Hz control step, the pipeline records hand commands, robot and vehicle states, CTB pose, and an opposing-contact indicator; contact requires fingertip contact on both sides of the array.Source: B. ISS Simulation · p. 3
- The cargo-motion metric measures CTB translation relative to Astrobee during confirmed contact, removing the vehicle's commanded translation and rotation.The reference is the CTB position in the Astrobee body frame at first confirmed contact.Source: B. ISS Simulation · p. 3
III. RESULTS
All ten live teleoperation trials maintained a continuous interval of confirmed opposing contact and completed the intended acquisition, translation, and right turn. Cargo motion, travel distance, and teleoperation duration varied across trials.
III. RESULTS
- All 10 trials established continuous opposing CTB contact and completed the intended acquisition, translation, and right-turn maneuver.The authors report that the integrated hand and base controller retained simulated cargo while the operator changed the free-flyer's trajectory.Source: §III · p. 3
- 35.6 mm was the median RMS CTB deviation from its first-contact Astrobee body-frame position; median peak deviation was 67.7 mm.The corresponding ranges were 16.4–79.0 mm RMS and 26.8–176.6 mm peak.Source: §III · p. 3
- 77.5 s was the median recorded teleoperation duration, and 9.30 m was the median net world-frame CTB displacement.The corresponding ranges were 67.3–97.5 s and 7.87–9.46 m; the confirmed-contact interval lasted a median 75.1 s.Source: §III · p. 3
IV. CONCLUSION & FUTURE WORK
The paper presents a deployable four-finger payload and bimanual teleoperation pipeline, demonstrating its use for cargo transport in an ISS-like simulation. Future work includes hardware characterization and transfer, broader evaluations, collision-aware control, and use of recorded trajectories for imitation-learning studies.
IV. CONCLUSION & FUTURE WORK
- The ISS-like simulation demonstrated teleoperated Astrobee cargo transport with the deployable four-finger payload and bimanual pipeline.The VR interface maps hand articulation to two finger pairs, wrist separation to the payload rail, and common wrist motion to the free-flying base.Source: §IV · p. 3
- Future work will characterize the completed payload hardware, evaluate repeated operators and tasks, and incorporate collision-aware Astrobee control.The pipeline is also slated for transfer to hardware.Source: §IV · p. 4
- Recorded teleoperated trajectories are proposed as demonstrations for subsequent imitation-learning studies using the same embodiment.Source: §IV · p. 4