Meetings

Link to page

5 June -Meeting

⁠

⁠

1) Anlayze all ROS bag files:

Ros bag info:

Ros bag info:

Subject

Session

File Name

Size

Duration

Messages

Compression

Sequence Duplicate

Dropped Frame

S1S1_2024-05-17-merged.bag

154.9 GB

9:44s (584s)

194,119

None

S02S01S1_2024-05-21-merged.bag

151.8 GB

11:45s (705s)

192,574

None

S02S02S1_2024-05-21-merged.bag

76.1 GB

3:59s (239s)

96,448

None

S03S01S1_2024-05-21-merged.bag

192.0 GB

16:20s (980s)

226,185

None

S04S01S1_2024-06-04-13-06-13.bag

126.3 GB

10:35s (635s)

36,555

None

Average

140.2 GB

10:28 (628s)

149,176

There are no rows in this table

⁠

Ros bag topics and types:

Ros bag topics and types:

Topics

Types

/cmd_vel/joystick

geometry_msgs/TwistStamped

/eeg_data

bcv/EEGData

/managed/joy

geometry_msgs/Twist

/odom

nav_msgs/Odometry

/processed_image

sensor_msgs/Image

/robot_status

std_msgs/Float32MultiArray

/status

ds4_driver/Status

/tf_static

tf2_msgs/TFMessage

/yoctopuce/fix

sensor_msgs/NavSatFix

/yoctopuce/imu

sensor_msgs/Imu

/yoctopuce/light

sensor_msgs/Illuminance

/zed/zed_node/left/camera_info

sensor_msgs/CameraInfo

/zed/zed_node/left/image_rect_color

sensor_msgs/Image

/zed/zed_node/odom

nav_msgs/Odometry

/zed/zed_node/path_map

nav_msgs/Path

/zed/zed_node/path_odom

nav_msgs/Path

/zed/zed_node/pose

geometry_msgs/PoseStamped

/zed/zed_node/pose/status

zed_interfaces/PosTrackStatus

/zed/zed_node/pose_with_covariance

geometry_msgs/PoseWithCovarianceStamped

/zed/zed_node/right/camera_info

sensor_msgs/CameraInfo

/zed/zed_node/right/image_rect_color

sensor_msgs/Image

There are no rows in this table

⁠

2) Check dropped messages: ??

3) Extract EEG data from each ROS bag file:

⁠

S1S1-eeg_data.txt

3.9 MB

⁠

S2S2-eeg_data.txt

1.6 MB

⁠

S4S1-eeg_data.txt

1.8 MB

⁠

4) Extract button pressed labels from each ROS bag file:

⁠

S1S1-joystick_data.txt

1.8 MB

⁠

S2S2-joystick_data.txt

910.6 kB

⁠

S4S1-joystick_data.txt

243.9 kB

⁠

5) Extract /processed image from each ROS bag file:

S1S1-PI-Video.avi

S2S2-PI-Video.avi

S4S1-PI-Video.avi

6) Extract /left image from each ROS bag file:

S1S1-LI-Video.avi

S2S2-LI-Video.avi

S4S1-LI-Video.avi

7) Filter all EGG data files:

⁠

S1S1-filtered-eeg_data.txt

3.3 MB

⁠

S2S2-filtered-eeg_data.txt

1.6 MB

⁠

8) Visualize EGG data files:

S1S1:

⁠

S4S1:

9) and 10) Clean EGG data files and merge with labeled joystick:

11) Apply models:

Model 1: EEGNet

Application: General EEG classification and Brain-Computer Interface (BCI) tasks.

Paper/GitHub:

EEGNet: A Compact Convolutional Network for EEG-based Brain-Computer Interfaces,

https://arxiv.org/abs/1611.08024⁠

⁠

https://github.com/aliasvishnu/EEGNet⁠

⁠

Model Description:

Layers:

Initial Convolutional Layer: 1 layer

Depthwise Convolutional Layer: 1 layer

Separable Convolutional Layer: 1 layer

Classification Layer: 1 layer

⁠

⁠

Justification: Designed to be compact and efficient for EEG data processing, capturing both spatial and temporal features.

Types: Conv2D, DepthwiseConv2D, SeparableConv2D, Linear

Loss Function: Cross-Entropy Loss

Epochs: 150

Metrics: Accuracy, Precision, Recall, F1-Score

Performance: Achieves high accuracy on multiple BCI datasets, often above 90% for binary classification tasks, 60-80% for multiclass classification

S1S1:

Accuracy: 53.51% Precision: 0.38 Recall: 0.54 F1-score: 0.44

S4S1:

Model 2: DeepConvNet

⁠

⁠

Application: General EEG tasks, more complex and deeper representations.

Paper/GitHub:

Deep learning with convolutional neural networks for EEG decoding and visualization,

https://arxiv.org/pdf/1703.05051⁠

⁠

https://github.com/robintibor/braindecode⁠

⁠

Model Description:

Layers:

Several convolutional layers followed by fully connected layers.

Typically includes 4 Conv2D layers and 2 fully connected layers.

Justification: Deeper layers allow for capturing more complex patterns in EEG data.

Types: Conv2D, MaxPooling, Dense (Fully Connected)

Loss Function: Cross-Entropy Loss

Epochs: 100

Metrics: Accuracy, Precision, Recall, F1-Score

Performance: Known for high accuracy, robustness across different EEG tasks.

S1S1:

Accuracy: 83.27% Precision: 0.84 Recall: 0.83 F1-score: 0.83

S4S1:

Model3: EEG-Transformer by eeyhsong

Description: Here applies a Transformer (ViT) model to 2-D physiological signal (EEG) classification tasks, which can be adapted for various tasks including driving intention prediction.

GitHub Repository:

https://github.com/eeyhsong/EEG-Transformer⁠

⁠

Justification:

Utilizes the attention mechanism to enhance spatial and temporal features.

Applies common spatial pattern (CSP) for feature enhancement.

Performance: The repository mentions achieving state-of-the-art performance in multi-classification of EEG signals.

S1S1:

Accuracy: 72.37% Precision: 0.84 Recall: 0.83 F1-score: 0.83

S4S1:

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.