Downloads

NeuTouch
Event-based Camera Settings
Code
Container/Weight Classification Dataset
YCB Dataset
Slip Classification Dataset
3D CAD Models

NeuTouch taxel information:

https://www.dropbox.com/scl/fi/vgrh30h6ir0r1fp2p4zjp/neutouch_taxel_info.zip?rlkey=522p47x2vhaevyyxd1w3ldk4l&dl=0
Information about NeuTouch’s taxel coordinates in a 2D frame.

Event-based Camera Settings:

https://www.dropbox.com/scl/fi/n186px1gz44drtmekgak8/prophesee.bias?rlkey=81p0ibtlc342e6whpmt04x4ej&dl=0
Hardware bias settings of Prophesee Onboard.

Replicating methods and results in our paper:

https://github.com/clear-nus/VT_SNN
Detailed documentation for installation and replicating results is on README.md.

Container/Weight Classification Dataset (Version 2.0)

Downloads

Lite (7G)
https://www.dropbox.com/scl/fi/5i84n61yr4zlri2o1kdzw/object_weight_lite.zip?rlkey=3vtzt6d4rqqgsnlrgpmtd56vp&dl=0
All data excluding 2D-RGB images from the 2 Intel RealSense D435s cameras.
Detailed README.txt provided.

RGB Preview (108M)
https://www.dropbox.com/scl/fi/t5afi02qx195di1jos3is/object_weight_rgb_preview.zip?rlkey=uvgp3566xgne28osz7cdy009a&dl=0
One example of 2D-RGB images from the 2 Intel RealSense D435s cameras.

RGB (86G)
Full 2D-RGB images from the 2 Intel RealSense D435s cameras. (warning: huge!)

Note: the data is splitted into 9 parts. Download all of them to be able to use the data. Due to the our dropbox policy for our data, you may be able to download only upto 50 GB data per day.

Preprocessing (321M)
https://www.dropbox.com/scl/fi/54sp0tpyx3c3y9dwvt4ca/object_weight_preprocessed.zip?rlkey=ubxs8pj5ujy1hy4egacu0vd3y&dl=0
Preprocessed ycb data, specific to our methods in our code.
README.txt provided.

Description

We obtain recordings from 4 distinct objects, each with 5 different weight classes. We have 40 samples per weight class.

We adopt the following naming convention for the individual classes:

107-a_pepsi_bottle:     0% Pepsi, 28g
107-b_pepsi_bottle:     25% Pepsi, 128g
107-c_pepsi_bottle:     50% Pepsi, 228g
107-d_pepsi_bottle:     75% Pepsi, 328g
107-e_pepsi_bottle:     100% Pepsi, 428g
108-a_tuna_fish_can:    0% Tuna, 29g
108-b_tuna_fish_can:    25% Tuna, 64g
108-c_tuna_fish_can:    50% Tuna, 99g
108-d_tuna_fish_can:    75% Tuna, 134g
108-e_tuna_fish_can:    100% Tuna, 169g
109-a_soymilk:          0% Soy, 26g
109-b_soymilk:          25% Soy, 101g
109-c_soymilk:          50% Soy, 176g
109-d_soymilk:          75% Soy, 251gIf you want the original version 1.0 of the dataset (Tactile data only), please download it here
109-e_soymilk:          100% Soy, 326g
110-a_coffee_can:       0% Coffee, 25g
110-b_coffee_can:       25% Coffee, 88g
110-c_coffee_can:       50% Coffee, 150g
110-d_coffee_can:       75% Coffee, 213g
110-e_coffee_can:       100% Coffee, 275g

The data we use for modelling are from the following folders:

aces_recordings: spiking tactile from NeuTouch tactile sensors.
prophesee_recordings: spiking vision from Prophesee Onboard.
traj_start_ends: timestamps for each phase of recording.

We also collect data from other sensors which are not used for modelling, but used for sanity checks:

2d_color_1: 2D RGB images from Intel RealSense D435, placed ~50cm away.
2d_color_2: 2D RGB images from Intel RealSense D435, mounted on Panda arm end-effector.
franka_states: Proprioceptive data from Franka Emika Panda.
rbtq_states: Proprioceptive data from Robotiq 2F-140.

Detailed documentation with respect to each modality is provided in README.txt.

If you want the original version 1.0 of the dataset (Tactile data only), please download it here (README.txt provided, 45MB):

YCB Dataset

Downloads

Lite (9G)
https://www.dropbox.com/scl/fi/awat9xihuq0sew0t1cmxy/ycb_lite.zip?rlkey=dmd1dmw2qazx8sgfesmufbavq&dl=0
All data excluding 2D-RGB images from the 2 Intel RealSense D435s cameras.
Detailed README.txt provided.

RGB Preview (111M)
https://www.dropbox.com/scl/fi/75bsxm2aa8b5nkc5g4k66/ycb_rgb_preview.zip?rlkey=1dw6pa4s7la84fjb9xsa5kzjg&dl=0
One example of 2D-RGB images from the 2 Intel RealSense D435s cameras.

RGB (100G)
Full 2D-RGB images from the 2 Intel RealSense D435s cameras. (warning: huge!)

Note: the data is splitted into 11 parts. Download all of them to be able to use the data. Due to the our dropbox policy for our data, you may be able to download only upto 50 GB data per day.

Preprocessing (500M)
https://www.dropbox.com/scl/fi/q48j0walio9jsokv1q46f/ycb_preprocessed.zip?rlkey=0jmbauv6q9e12tgurdssltulh&dl=0
Preprocessed ycb data, specific to our methods in our code.
README.txt provided.

Description

We obtain recordings from 36 different objects from YCB dataset. There are 25 samples per object.

We use YCB’s naming convention for objects:

003_cracker_box
004_sugar_box
006_mustard_bottle
010_potted_meat_can
040_large_marker
041_small_marker
046_plastic_bolt
061_foam_brick
065-a_cups
065-b_cups
065-c_cups
065-d_cups
065-e_cups
071_nine_hole_peg_test
072-d_toy_airplane
072-f_toy_airplane
072-i_toy_airplane
072-j_toy_airplane
073-a_lego_duplo
073-d_lego_duplo
073-h_lego_duplo
073-n_lego_duplo
073-o_lego_duplo
073-p_lego_duplo
077_rubiks_cube
101_rope
102-a_duck
102-b_duck
103-a_evian
103-b_evian
103-c_evian
104_sponge
106-a_vitasoy
106-b_vitasoy
107-a_pepsi_bottle
107-b_pepsi_bottle

The data we use for modelling are from the following folders:

aces_recordings: spiking tactile from NeuTouch tactile sensors.
prophesee_recordings: spiking vision from Prophesee Onboard.
traj_start_ends: timestamps for each phase of recording.

We also collect data from other sensors which are not used for modelling, but used for sanity checks:

2d_color_1: 2D RGB images from Intel RealSense D435, placed ~50cm away.
2d_color_2: 2D RGB images from Intel RealSense D435, mounted on Panda arm end-effector.
franka_states: Proprioceptive data from Franka Emika Panda.
rbtq_states: Proprioceptive data from Robotiq 2F-140.

Detailed documentation with respect to each modality is provided in README.txt.

Slip Classification Dataset

Downloads

Lite (300M)
https://www.dropbox.com/scl/fi/dko6knktsz22orwqi19o0/slip_classification_lite.zip?rlkey=47kgm0uehfgnsg844oasfw31k&dl=0
All data excluding 2D-RGB images from the 2 Intel RealSense D435s cameras.
Detailed README.txt provided.

Full (4.4G)
https://www.dropbox.com/scl/fi/3m5u3u5zdkpc1bcon084v/slip_classification_full.zip?rlkey=xhqircf6cfw1erun20255z6gh&dl=0
All data.
Detailed README.txt provided.

Preprocessing
https://www.dropbox.com/scl/fi/jn16aqx7qkhbm80agcw0g/slip_preprocessed.zip?rlkey=j7s8piwvhgd4mvmfxaqiiqmpc&dl=0
Preprocessed slip data, specific to our methods in our code.
README.txt provided.

Description

We collect 50 recordings with rotational slip rotate_01 to rotate_50, and 50 without slip stable_01 to stable_50.

The data we use for modelling are from the following folders:

aces_recordings: spiking tactile from NeuTouch tactile sensors.
prophesee_recordings: spiking vision from Prophesee Onboard.
traj_start_ends: timestamps for each phase of recording.

We also collect data from other sensors which are not used for modelling, but used for sanity checks:

2d_color_1: 2D RGB images from Intel RealSense D435, placed ~50cm away.
2d_color_2: 2D RGB images from Intel RealSense D435, mounted on Panda arm end-effector.
franka_states: Proprioceptive data from Franka Emika Panda.
rbtq_states: Proprioceptive data from Robotiq 2F-140.
optit_recordings: Pose of end effector and object, from OptiTrack, and also heuristics-based checks on when object is lifted off table or when object first rotates.

Detailed documentation with respect to each modality is provided in README.txt.

3D CAD Models:

https://www.dropbox.com/scl/fi/vzzqhydl9l0f6yknctfz2/3D_print_parts.zip?rlkey=sfgvy7vhv9warg8an1u30rkh0&dl=0
STL files for 3D CAD models to attach Neutouch, event-based and rgb cameras to Franka Emika Panda arm. README.txt provided.

Home Downloads FAQ

Downloads

Table of Contents

NeuTouch taxel information:

Event-based Camera Settings:

Replicating methods and results in our paper:

Container/Weight Classification Dataset (Version 2.0)

Downloads

Description

YCB Dataset

Downloads

Description

Slip Classification Dataset

Downloads

Description

3D CAD Models: