Source-linked AI summary
Robot Aware Computational Design of Object Specific Passive Grippers for Additive Manufacturing
Abdullah Yahya Abdullah Omaisan, Ibrahim Sheikh Mohamed
TL;DR
Object-specific passive gripper design couples contact, retention, robot insertion, and structural safety, choices that additive manufacturing alone does not make. The paper combines these steps in a traceable computational pipeline; four archived designs pass recorded digital gates, but remain short of operational release pending physical validation.
Problem
Additive manufacturing does not resolve where a gripper should contact an object, which passive retention mechanism to use, whether the robot can insert it, or whether its printed load path is safe.
Source: §1 · p. 2Method
The pipeline joins robot-aware contact screening, selection among six passive mechanisms, conformal geometry, insertion motion, directional finite-element analysis, and protected topology optimization.
Source: §1 · p. 2Results
All four archived designs cleared the recorded nominal digital gates; under the wider pose box, A1–A4 retained 134, 31, 81, and 29 of 160 trials, respectively.
Source: §4.2 · p. 9, §6 · p. 11Takeaways & Limitations
The four object-specific records support computational screening, but all remain prototypes until calibrated material and physical fit, load, shake, and cycle tests pass.
Source: §6 · p. 11Nominal material properties and absent coupon-calibrated, instrumented fit, load, shake, and cycle tests keep every release at digital screening rather than operational release.
Source: §4.2 · p. 10
Abstract
from arXiv · showhide
This paper presents an end-to-end computational pipeline that converts a selected object mesh, a measured object state, and a selected six-axis robot into an object-specific, unactuated, additively manufacturable gripper. The method couples exact-mesh RGB-D/ICP pose registration, deterministic surface-contact sampling, uncertainty-aware wrench screening, selection among six passive capture mechanisms, object-conformal surface synthesis, full-orientation robot inverse kinematics, a swept-volume-aware manufacturing domain, directional fused-deposition finite-element screening, and constrained three-dimensional SIMP topology optimization. Unlike workflows that treat grasp selection, tool geometry, motion, and structural design as separate problems, every exported design is bound to the source mesh, object pose, robot flange, contact set, and insertion hypothesis by a traceable design identifier. We derive the implemented registration, contact, fit-tolerance, finite-element, and density-optimization equations and prove three properties of the numerical construction: nodal load preservation, monotonic compliance sensitivity under SIMP interpolation, and voxel-domain containment after topology post-processing. Four archived object-specific attempts - a rabbit, camera flange, 3DBenchy, and faceted bust - meet the nominal fit, uncertain-wrench, runtime-sweep, and baseline/post-topology FEA gates. A deliberately enlarged +/-3 mm, +/-5 degree pose stress check differentiates the designs, retaining 29-134 of 160 simulated trials. Their reconstructed topologies retain 92.0-97.6% of the FE domain because functional regions are protected. Archived robot photographs show the corresponding printed assemblies qualitatively, while nominal material properties and absent coupon-calibrated, instrumented tests keep all four at digital-screening status rather than operational release.
1 Introduction
Object-specific passive grippers require coordinated choices about contact, retention, robot insertion, and structural safety that additive manufacturing alone does not resolve. The framework joins those decisions in a traceable design process and package.
1 Introduction
- Additive manufacturing does not determine contact, passive retention, robot insertion, or print safety, while passive grippers must achieve insertion, locking, carrying, and release without a gripper-side actuator.The paper identifies these coupled choices as a reason object-specific grippers are commonly produced through manual, experience-dependent iteration.Source: §1 · p. 2
- The framework converts an object mesh, pose and mass, robot and manufacturing profiles, and a pick/place task into a traceable design package.The package includes ranked contact evidence, a connected watertight preform, an insertion–lock–release program, robustness records, finite-element reports, and an optional optimized STL.Source: §1 · p. 2
- Its principal contributions join robot-aware grasp and mechanism design with directional structural screening, protected topology optimization, and four traceable object-specific design records.The numerical construction includes proofs of force preservation, compliance monotonicity, and voxel-domain containment.Source: §1 · p. 2
2 Related Work
Prior work automates fingers, grasp proposals, registration, and structural refinement, while the paper positions its contribution as an auditable pipeline for selecting and carrying through an explicit passive mechanism. Its design domain fixes motion, fit, interfaces, and other functional constraints before material removal.
2 Related Work
- Automated finger and grasp methods address contact or grasp proposals, but do not produce a flange-connected passive body and its insertion program; the paper positions its pipeline as complementary to the passive baseline of Kodnongbua et al..Prior examples include printable contact primitives, conformal caging pads [4], learned finger volumes, and learned grasp quality.Source: §2 · p. 2
- The pipeline's contribution is an auditable process that selects an explicit passive mechanism, binds it to a robot, flange, and camera configuration, and screens conformal contacts and structural design.The screening includes uncertainty and structural analysis.Source: §2 · p. 3
- Selected-mesh registration follows Besl and McKay's iterative closest point formulation, with the accepted full pose and exact mesh identity passed into trajectory planning.Source: §2 · p. 3
- Contact ranking draws on geometric restraint, force closure, and wrench-space quality, then separates ideal-set wrench screening from deterministic perturbation audits informed by uncertainty sampling.Neither score is presented as a measured grasp probability because the manufactured body also relies on distributed surfaces and positive retention.Source: §2 · p. 3
- Structural refinement builds on SIMP and finite-element sensitivity, morphology filters, and direction-dependent FDM properties, but fixes motion, fit, flange, camera clearance, stops, walls, and skin before material removal.The paper also distinguishes this restricted domain from prior topology-optimization applications to fingers and grippers and a food-serving robot.Source: §2 · p. 3
3 Method
The method links object and robot inputs to a traceable gripper design through ordered contact, geometry, motion, and structural checks. It combines passive-mechanism synthesis with robot-aware fit and manufacturing-aware structural screening.
3.1 Formulation, Mesh Audit, and Contacts
- The design is a sequence of ordered feasibility gates, with a SHA-256 identifier binding each artifact to its mesh, pose, robot, contacts, mechanism, motion, clearances, and voxel settings.Contact candidates come from deterministic surface sampling and are screened for spacing, projected area, normal opposition, blocked paths, and uncertain-wrench limits.Source: §3.1 · pp. 3–4
3.2 Mechanism and Geometry Synthesis
- Mesh descriptors select an initial passive capture family, while conformal patches join family features, a flange mount, load paths, and a camera-safe neck into a checked preform.Voxel operations subtract the object, insertion sweep, mount holes, and keep-outs; the resulting geometry is checked for connectivity, retained fit zones, wall thickness, watertightness, and clearance.Source: §3.2 · pp. 4–5
3.3 Selected-Mesh Registration, Motion, and Fit Robustness
- Selected-STL samples are aligned to segmented depth data by robust six-DoF point-to-plane ICP, with fitness, RMSE, and correction thresholds gating registration.Seven local yaw seeds initialize ICP at progressively finer voxel and correspondence scales; changing the selected path or digest invalidates the pose.Source: §3.3 · p. 5
- The registered full orientation informs insertion and inverse-kinematics planning, while fit robustness is tested over 160 seeded trials spanning translation and rotation perturbations.The stored program covers home through insertion, locking, lifting, transfer, placement, unlocking, and retraction; archived runs also include a 233-frame runtime sweep.Source: §3.3 · pp. 5–6
3.4 Directional FEA and Protected Topology Optimization
- Structural screening uses distributed contact loads, trilinear eight-node hexahedra, and nominal transversely orthotropic PLA properties; its stress check is a conservative digital screen, not physical certification.The model applies a 7.5mg load and uses stated XY/Z allowable stresses and a 1 mm displacement limit.Source: §3.4 · pp. 6–7
- The numerical construction proves that nodal transfer preserves applied resultants and that increasing element density cannot increase compliance under the stated stiffness condition.The compliance result follows from a nonpositive density derivative when free-DOF stiffness is positive-definite.Source: §3.4 · pp. 6–7
- The SIMP formulation constrains density while protecting the mount, neck, contacts, stops, functional shell, and outer skin through filtering and post-processing.The workflow applies thresholding, void filtering, mount-component restoration, and post-topology FEA.Source: §3.4 · p. 7
- Post-processed topology voxels remain within the collision-approved preform because reconstruction operations remove material or intersect grown voxels with that domain.Surface extraction still requires a separate triangle-mesh collision audit.Source: §3.4 · p. 7
4 Experimental Setup and Results
Four archived design cases were assessed through enlarged pose-stress audits and structural re-analysis, with results bounded to digital screening rather than physical reliability.
4.1 Dataset and Protocol
- The four archived attempts are documented design cases, not a statistical estimate of success over an object population.Figure 5 links each object to its selected mechanism and shows computational assemblies, pre-pickup robot poses, and printed grippers holding objects; the photos are qualitative, not instrumented validation.Source: §4.1 · pp. 7–8, Figure 5 · p. 8
- All four attempts used a 0.20 kg declared load, a DOBOT CR10A, nominal PLA properties, and common contact-search, fit-trial, and wrench-uncertainty settings.The geometry grid was 0.55 mm and the FE elements were approximately 3 mm; all optimizations converged in eight iterations.Source: §4.1 · pp. 8–9
- The implemented ICP pathway was not evaluated as a pose-accuracy experiment.Thus, the archived attempts do not establish measured pose accuracy.Source: §4.1 · p. 9
4.2 End-to-End Outcomes
- A1–A4 retained 134, 31, 81, and 29 of 160 trials, respectively, in the enlarged pose box; all cleared runtime-sweep, FEA, and watertight-reconstruction gates.The gates are bounded digital checks, not estimates of physical reliability.Source: §4.2 · p. 9
- The finite runtime sweep used a capsule-plus-visual-bounds collision model, while nominal material properties and absent coupon-calibrated, instrumented tests leave physical validation pending.Controller tracking, calibration uncertainty, and cable motion also remain unevaluated.Source: §4.2 · pp. 9–10
4.3 Structural Evidence
- Across A1–A4, maximum von Mises stress changed from 0.501–1.537 MPa before optimization to 0.501–1.647 MPa after reconstruction, while maximum displacement changed from 0.061–0.914 mm to 0.061–0.991 mm.Final FE material fractions were 92.0–97.6%, above the nominal 0.82 target because functional regions are protected; these results do not establish general mass savings or physical strength.Source: §4.3 · p. 10
5 Discussion and Limitations
The four attempts support staged co-design: screening identifies promising contacts, while conformal geometry, robot motion, and fixed functional regions shape usable and structurally screened tools. The paper also identifies absent physical validation and model limitations, with calibration and experiments as next steps.
5 Discussion and Limitations
- Protected fit surfaces, flange, motion corridor, and keep-outs should be fixed before structural optimization; target density does not promise mass reduction in compact grippers.Exported volume and post-FEA response are the relevant outputs.Source: §5 · p. 10
- The four designs lack physical validation, and nominal linear orthotropic FEA omits coupon calibration, nonlinear contact, creep, fatigue, bolt preload, and print defects.The robot sweep also leaves controller tracking, calibration uncertainty, and cable motion unevaluated; ICP lacks a calibrated pose-error dataset.Source: §5 · p. 10
- Next steps are coupon-calibrated anisotropic models, physical fit/load/shake/cycle tests, and calibrated triangle-mesh robot collision checks.Measured dimensional and force data could replace uniform fit allowances and calibrate wrench-residual and zone-margin relations to holding force.Source: §5 · p. 10
6 Conclusion
The conclusion presents a traceable pipeline and four archived designs that pass recorded digital gates. It emphasizes that they remain prototypes pending calibrated material and physical testing.
6 Conclusion
- The framework joins contact screening, passive mechanisms, conformal geometry, robot-aware insertion, directional H8 finite elements, and protected SIMP topology optimization in one traceable process.The analysis proves preservation of applied resultant force, monotonic compliance sensitivity, and containment of post-processed topology voxels.Source: §6 · p. 11
- Four archived dashboard attempts connect each selected object and mechanism to computational fit, wrench, runtime-sweep, FEA, topology, and release evidence.All four pass the recorded digital gates.Source: §6 · p. 11
- The four designs remain prototypes until calibrated material and physical fit, load, shake, and cycle tests pass.Source: §6 · p. 11
Declarations
The declarations state that ethics approval and consent are not applicable because the study involved no human participants, human data, animals, or biological materials.
Declarations
- Ethics approval and consent were not applicable because the computational engineering study involved no human participants, human data, animals, or biological materials.Source: Declarations · p. 11