Component Testing & Experiment Results

Perception

The following table shows the tests performed to test the cornstalk width detection.

Test PerformedMetricConclusion
Depth Movement± 0.45 mmGood for distances  > 10 cm
Horizontal Movement± 3.65 mmMinor difference due to image distortion
Yaw Angle± 1.20 mmPerformed as expected
Roll AngleN/AStalk not detected at higher angles
Width± 1 mmPerformed as expected
Ovular MeasurementN/APerformed as expected
OcclusionN/AOccluded regions detected with lower conf.

Manipulation

Test Performed# TestsGrasp RegionSuccessConclusion
Approach and Sweep Cornstalks20y > 0.6
x > 0.2
x <= 0.2
100%xArm performed as expected
Approach and Sweep Cornstalks10y < 0.6
x > 0.2
x <= 0.2
80%xArm workspace is very limited with corn rows closer to base. Singularities faced in planning.
Hook Cornstalks20y > 0.6
x > 0.2
x <= 0.2
90%xArm performed as expected
Hook Cornstalks10y < 0.6:
x > 0.2
x <= 0.2

90%xArm workspace is very limited with corn rows closer to base. Singularities faced in planning.

End Effector

Tests Performed# TestsSuccessConclusion
Repeatability8100%Gripper is reliable
Varying DIA8100%Gripper can handle 17-45 mm DIA cornstalks effectively
Varying Orientation7100%Sensor insertion is not highly dependent on stalk orientation
Varying X-Offset30%Need simple visual servoing
Varying Y-Offset4100%Insertion quality not dependent on Y-offset