How We Test

Automated validation pipeline powered by NVIDIA Isaac Sim

Policy Upload
SHA256 Verify
Isaac Sim Physics
68 Scenarios
14 Metrics
PASS / FAIL

Parameter Space Design

Why we chose these parameters — each backed by peer-reviewed research

ParameterRangeScalePaperWhy
friction0.05 – 1.2log-uniformSIMPLER 2024Friction-success correlation r=+0.36 (#1)
mass0.05 – 2.0 kglog-uniformSIMPLER 2024Reflects Franka payload limits
com_offset0.0 – 0.4uniformSuction Grasp 2025CoM bias → grasp stability shift
size0.02 – 0.12 muniformSIMPLER 2024Failure rate spikes below 4cm
ik_noise0.0 – 0.04 raduniformICRA Sim2Real 2025Sim-to-Real control error simulation
obstacles0 – 4integerRoboFAC 2025Collision rate doubles at 3+ obstacles
shape5 typescategoricalGrasp Anything 2024box, cylinder, sphere, L, irregular
placement14 typescategoricalALEAS 2025rotation / tilt / edge placements

Two-Stage Adaptive Sampling: Stage 1 uniform LHS 20K + Stage 2 boundary-focused LHS 10K → AUC 0.65 → 0.777 (+19.5%)

Failure Taxonomy

6 failure types based on RoboFAC (2025)

Grasp Miss

Gripper approached but failed to grasp the object

size < 3cm, approach_angle > 60°

Collision

Robot collided with table, object, or obstacle

obstacles ≥ 3, cluttered scene

Drop / Grip Loss

Object dropped during transport after grasp

friction < 0.2, mass > 1.5kg

Timeout

Task not completed within time limit

complex IK solution, non-vertical approach

Slip Event

Slippage detected during grasp (object not dropped)

friction < 0.3, com_offset > 0.2

Position Deviation

Object not placed accurately at target position

reach_ratio > 0.82, IK noise

Cross-Robot Validation

Same experiments on two robots → discover universal danger zones

Franka Panda (7DOF)

20,000

experiments · SR 48.6% (uniform 10K + boundary 10K)

danger zone: 7,808

UR5e (6DOF)

10,000

experiments · 74.3% success rate

UR5e PickPlaceController + SurfaceGripper (suction)

danger zone: 2,570

UR3e (6DOF)

10,000

Lightweight robot validation

NEW

UR10e (6DOF)

10,000

Heavy-duty robot validation

NEW

Universal danger zone: mass > 0.93 kg → both robots SR < 40%. Boundary equation: μ*(m) = (1.469 + 0.419m) / (3.691 - 1.400m)

Statistical Confidence

Wilson Score Interval — finite sample confidence bounds (SureSim 2025)

50,000+

Samples

±0.6%

95% CI margin

p < 0.001

Statistical significance

Wilson Score: p̂ ± z·√(p̂(1-p̂)/n + z²/4n²) / (1 + z²/n)

n=50,000, p̂=0.557, z=1.96 → CI = [0.553, 0.561]

References

Papers that RoboGate is built on

SIMPLERCoRL 2024

friction × mass joint sampling, Sim-to-Real gap quantification (24-30%)

ALEASRSS Workshop 2025

Latin Hypercube Sampling — 2-3× space coverage vs. random

ICRA Sim2RealICRA 2025

IK noise injection for domain randomization

SureSimL4DC 2025

Wilson Score Interval — finite sample confidence bounds

RoboFACarXiv 2025

6-type failure classification taxonomy

Suction GraspRA-L 2025

Center of Mass offset effect on grasp stability

Learning to Grasp AnythingCoRL 2024

5 object shapes (box, cylinder, sphere, L, irregular)

Isaac LabNVIDIA 2025

GPU parallel environments (4096 envs), Newton Physics engine