A robotic fingertip spinning a pen at high speed, with blue motion trails

Mastering “Pen” Spinning in Diverse Styles

Robotic pen spinning

Main Real World Results (featured by high-speed, fine-grained movements, sophisiticated gaiting, and realiability)

Diverse “Pen” Spinning Skills

Pass-v0

01 Reliable one-episode success

100% (50 in 50) success rate

02 Sustained rotation (averaged 4.025 episodes across 20 evaluations)

Avg. 4.025 episodes, Total 80.5 episodes, #Evaluations 20

01 Sustained rotation (all in the original 1x speed)

Four examples

01 Sustained rotation (all in the original 1x speed)

Four examples

Pass-v1

01 Reliable one-episode success (48 in 50)

96% (48 in 50) success rate

02 Sustained rotation (4 rotation episodes)

4 rotation episodes

Pass-v2

01 Reliable one-episode success (45 in 50)

90% (45 in 50) success rate

02 Sustained rotation (2 rotation episodes)

2 rotation episodes

Pass-v3

01 Reliable one-episode success (48 in 50)

96% (48 in 50) success rate

02 Sustained rotation (3 rotation episodes)

3 rotation episodes

Thumb-Around

01 Reliable one-episode success (49 in 50)

98% (49 in 50) success rate

04 Diverse Thumb-Around Spinning

Diverse Thumb-Around Spinning Clips

Sonic (down)

01 Reliable one-peisode successes (50 in 50)

100% (50 in 50) success rate

04 Sonic Spinning

Sonic Spinning

Triangle Pass (palm-up)

01 Reliable one-episode successes (50 in 50)

100% (50 in 50) success rate

02 Sustained Rotation (avg. 39.9 episodes across 5 trails)

Avg. 39.9 episodes

01 Reliable one-episode successes (50 in 50)

100% (50 in 50) success rate

02 Sustained Rotation (25+ mins, 91 episodes x 2pi)

25+ mins, 91 episodes x 2pi

01 Reliable one-episode successes (100 in 100)

100% (100 in 100) success rate

02 Sustained Rotation (6 mins)

6 mins

Triangle Pass (palm-down)

01 Reliable one-episode successes (50 in 50)

100% (50 in 50) success rate

02 Sustained Rotation (avg. 40.1 episodes across 5 trails)

Avg. 40.1 episodes (20 episodes (4.5mins) - 43 episodes (~13mins) - 27.5 episodes (8mins 26s) - 77 episodes (23mins 38s) - 32.5 episodes (10mins))

Triangle Pass (palm-down, reversed)

01 Reliable one-episode successes (50 in 50)

100% (50 in 50) success rate

02 Sustained Rotation (avg. 10.7 episodes across 5 trails)

Avg. 10.7 episodes

Four-Finger-Pass (palm-down)

01 Reliable one-episode successes (100 in 100) and Sustained Rotation (30+ mins, 134 x 2pi in total)

100% (100 in 100) success rate and 30+ mins, 134 x 2pi in total

02 Sustained Rotation (15+ mins)

15+ mins

Four-Finger-Pass (palm-down, LEAP)

01 Reliable one-episode successes (50 in 50)

100% (50 in 50) success rate

02 Reliable one-episode successes (48 in 50)

96% (48 in 50) success rate

03 Sustained Rotation

sustained rotation

04 Sustained Rotation (2 episodes)

2 episodes

Zero-shot Adaptation to 1) different object materials, 2) changed dynamics, 3) various object shapes.

Zero-shot Adaptation

01 Adapation to new geometry (cylinder -> hexagonal prism, octagonal prism; continuous rotation & multiple successes)

cylinder -> hexagonal prism, octagonal prism

02 Adapation to unseen materials (rigid -> soft; continuous rotation & multiple successe)

rigid -> soft

03 Adaptation to diverse shapes (T shape, L shape, ellipse, c-ellipse)

T shape, L shape, ellipse, c-ellipse

04 Remove the tape wrapped around the fingers (success rate: ~60%)

Low-friction scenario

05 Adaptation to new object shapes (T shape, L shape, long board)

T shape, L shape, long board

06 Adaptation to OOD object mass

Heavy object

06 Pass-v1 (cylinder -> octagonal prism)

cylinder -> octagonal prism

05 Adaptation to new object material (rigid -> soft)

rigid -> soft

05 Adaptation to new object geometry (T shape, L shape)

T shape, L shape

06 Adaptation to new object geometry (cylinder -> hexagonal prism)

cylinder -> hexagonal prism

09 Adaptation to new object geometry (cylinder -> octagonal prism)

cylinder -> octagonal prism

10 Adaptation to new object material (rigid -> soft)

rigid -> soft

09 Adaptation to new object geometry (cylinder -> hexagonal prism and octagonal prism) and new object material (rigid -> soft)

hexagonal prism, octagonal prism, soft material

01 A low-friction object (from lees than 50% success rate to 94% success rate (47/50))

A low-friction object

Some interesting results from our experiments.

Few-shot Adaptation

06 Adaptation to new object geometry (15cm x 2cm x 5cm, 16mins, from around 45% success rate to 100% success rate and sustained rotation; 15cm x 2cm x 10cm; 15cm x 2cm x 7cm )

Adaptation to new object geometry

02 A thinner ``pen''' (2.0 cm -> 1.4 cm, from less than 10% success rate to ~100% success rate)

Thick pen (2.0 cm) to thin pen (1.4cm)

03 Cuboid -> Octagonal Prism

Cuboid -> Octagonal Prism

04 Thumb-around (unssen materials, rigid -> soft; from less than 10% success rate to 96% success rate (48/50))

Adapation to unseen materials

05 New object geometry (from around 40% to 100% success rate)

New object geometry

06 Object with a larger diamerter (1.4 cm -> 1.6 cm, from around 50% success rate to 100% success rate)

Adapation to object with a larger diamerter

05 New object geometry (cubodi -> cylinder, from around 30% to 100% success rate)

New object geometry

05 Adaptation to new object shapes (cuboid -> octagonal prism, from around 50% to 100% success rate)

Adaptation to new object shapes (cuboid -> octagonal prism)

Robustness to Human Disturbances

Robustness to Human Disturbances

01 1) Push the object out of the balanced pose, 2) Drag the object to an extreme initial position

Three-finger-pass (palm-up)

02 Consecutive successes with disturbances (three-finger-pass, palm-up)

Three-finger-pass (palm-up)

03 1) Push the object, 2) Drag the object out of the balancing pose

pass-v0

04 Consecutive successes with disturbances (pass-v0)

pass-v0

05 Push the object off the balance x 3

three-finger-pass, palm-down

06 Consecutive successes with disturbances (three-finger-pass, palm-down)

three-finger-pass, palm-down

Full takeover and partial online takeover

Trails on Human Takeover for Failure Recovery

01 Pass-v1, full takeover

full takeover

02 Pass-v1, partial online takeover

Partial online takeover

01 Pass-v2, full takeover

full takeover

02 Pass-v2, partial online takeover

Partial online takeover

01 Triangle pass (palm-up), full takeover

full takeover

02 Triangle pass (palm-up), partial online takeover

Partial online takeover

Initial position distribution, rollout distribution

Analysis on Cross-Rollout Variations

Three-finger-pass (palm up)

01 Consecutive success one-episode rollout (59/59)

59 consecutive successes · real-world footage at 10× speed.

02Rollout distribution

Drag to orbit · scroll to zoom
59 recorded trajectories, overlaid in 3D. Pause, scrub time, or change playback speed.
Open standalone viewer ↗

03Initial position distribution

Rotate · zoom · select a point
One point per rollout, at frame 0. Coordinates in mm relative to one shared centroid.
Full report & XY / XZ / YZ views ↗

Pass-v0

01 Successful one episode rollouts (the video is play in 10x speed)

Pass-v0

02Rollout distribution

Drag to orbit · scroll to zoom
93 recorded rollouts with measured and target hand poses overlaid in 3D. Pause, scrub time, or change playback speed.
Open standalone viewer ↗

03Initial position & orientation

Rotate · zoom · select a point
93 initial poses, one per rollout at frame 0. Switch between XYZ positions and quaternion orientations.
Full report & XY / XZ / YZ views ↗

Three-finger-pass (palm-down, reversed)

01 Successful one episode rollouts (the video is play in 10x speed)

Three-finger-pass (palm-down, reversed)

02Rollout distribution

Drag to orbit · scroll to zoom
33 successful rollouts with measured and target hand poses. Unreliable object-tracking intervals are hidden.
Open standalone viewer ↗

03Initial position & orientation

Rotate · zoom · select a point
33 frame-zero poses · XYZ / Quaternion. The diamond marks one pose estimated across a brief tracking gap.
Full report & XY / XZ / YZ views ↗

Three-finger-pass (palm-down)

01 Successful one episode rollouts (the video is play in 10x speed)

Three-finger-pass (palm-down)

02Rollout distribution

Drag to orbit · scroll to zoom
41 successful rollouts with measured and target hand poses. Pause, scrub time, or change playback speed.
Open standalone viewer ↗

03Initial position & orientation

Rotate · zoom · select a point
41 initial poses, one per rollout at frame 0. Switch between XYZ positions and quaternion orientations.
Full report & XY / XZ / YZ views ↗

Pass-v1

01 Successful one episode rollouts (the video is play in 10x speed)

Pass-v1

02Rollout distribution

Drag to orbit · scroll to zoom
50 successful rollouts with measured and target hand poses. Pause, scrub time, or change playback speed.
Open standalone viewer ↗

03Initial position & orientation

Rotate · zoom · select a point
50 initial poses, one per rollout at frame 0. Switch between XYZ positions and quaternion orientations.
Full report & XY / XZ / YZ views ↗

Pass-v3

01 Successful one episode rollouts (the video is play in 10x speed)

Pass-v3

02Rollout distribution

Drag to orbit · scroll to zoom
59 successful rollouts with measured and target hand poses. Pause, scrub time, or change playback speed.
Open standalone viewer ↗

03Initial position & orientation

Rotate · zoom · select a point
59 initial poses, one per rollout at frame 0. Switch between XYZ positions and quaternion orientations.
Full report & XY / XZ / YZ views ↗

Fun Results

Fun Results

01 Fun Result (1) - Pass-v2

Fun Result (1) - Pass-v2

02 Fun Result (2) - Policy resulting from an intermediate reward searched by GPT

Fun Result (2)

The idea

TODO

The method

TODO