RoboSub 2024 (#555)

* tet

* Add color correction

* rename

* rename

* Add credit

* remove comment

* Lint

* Color correction

* 6 shaves

* revert

* fix

* always queue rgb

* color correct

* Add arg

* Add arg

* lint

* remove loginfo

* 🙃

* Tuned static power global

* Added RectInfo.msg

* Tuned static power global

* Switched usb_camera to bottom

* Added blue rectangle detector

* Added blue rectangle detector to sensors.launch

* Fixed move_to_cv_obj

* Added blue rectangle to CV interface

* Modified prequal task code

* Added prequal task to task runner

* Deleted rospy.loginfo statements from blue rectangle detector

* Added rect info subscriber to cv interface

* Modified prequal code

* Decreased angular tolerance

* Added depth to state interface

* Added blue rect touching top and touching bottom and outlier filtering to cv interface

* Updated prequal task

* Improved code to compute blue rectangle angle

* Final prequal code

* Added enable controls to controls interface

* Removed outlier filtering from blue rectangle in cv interface and added rect angle publisher

* Add maxsize and blocking

* Added gate style task

* First draft of buoy detector with conotour matching

* Changed usb camera channel back to 0

* Completed barrel roll task

* Removed unnecessary code

* Added IMU to state interface

* Added HSV filtering to buoy detector

* Improved buoy detector

* Added bbox filtering

* Completed buoy detector

* Buoy contour image

* Removed print

* Added buoy detector to sensors.launch

* First draft of buoy task

* Added buoy to cv interface

* Switched task runner to buoy task

* Added bbox related things to geometry_utils

* launch arg

* compile serial

* refactor to use open/close

* removed ros layer

* refactored arduino, launch

* constant message length

* updated docs, adds serial.launch alias

* Tightened HSV values for buoy detector and commented out filtering code

* Completed buoy task

* Copy models

* Add ros-yolov7

* Add ros-image-transport

* Patch bugs in ros-yolov7

* Add ros-yolov7 deps

* Add weights and labels

* Color based on class

* Add paths and tweak params

* Upload best.pt

* Changed cuda to cpu for yolov7

* Increased yolov7 confidence threshold to 0.90

* Corrected yolov7 labels order

* Reversed channel order in np_img_orig in yolov7

* Tuned PID

* Increased angular tolerance

* Added orig IMU to state interface

* Added gate detections to cv interface

* Added initial submerge, coin flip, and gate task to task runner with input

* Added initial submerge, coin flip, and gate task

* Tuned static power

* Separated angular tolerance into roll, pitch, yaw tolerances

* Added orig_state to state interface

* Updated CV with accurate gate image height and to use only object width for distance computation

* Added coin flip and gate style task to task runner

* Added keep_level to move to pos local

* Updates to gate tasks

* Updated thruster arduino config

* Removed stereo vision from spatial detection to improve detection rate

Co-authored-by: Maxwell Lin <maxwellmlin@users.noreply.github.com>

Co-authored-by: Vedarsh Shah <vedarshshah@users.noreply.github.com>

* Spatial detection fov increase, color fixes, switched to gate model

* add 5 and 10 shave count

* Reverted to ROS thrusters, added xyz calculation in depthai spatial detection, improvements to comp tasks

* Switched gate model to 5 shave

* Improved buoy detections

* Fixed depthai camera connect

* Publish rgb detections after filtering them

* Increase pitch tolerance to 0.2

* Added buoy distance computation

* Added buoy task to task runner

* Comp tasks updates

* Refactoring started

Co-authored-by: Maxwell Lin <maxwellmlin@users.noreply.github.com>
Co-authored-by: Rico Zhu <ricohasgithub@users.noreply.github.com>
Co-authored-by: Mathew Chu <chugames0414@gmail.com>
Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>
Co-authored-by: Raul Galvez <RaulGalvez288@users.noreply.github.com>
Co-authored-by: Nathan Yang <Knicey@users.noreply.github.com>

* continue reading from USB camera upon failure

* Refactoring 2

Co-authored-by: Maxwell Lin <maxwellmlin@users.noreply.github.com>
Co-authored-by: Rico Zhu <ricohasgithub@users.noreply.github.com>
Co-authored-by: Mathew Chu <chugames0414@gmail.com>
Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>
Co-authored-by: Raul Galvez <RaulGalvez288@users.noreply.github.com>
Co-authored-by: Nathan Yang <Knicey@users.noreply.github.com>
Co-authored-by: Amir Ergashev <amirergashev@users.noreply.github.com>
Co-authored-by: Hung Le <viethungle-vt1401@users.noreply.github.com>
Co-authored-by: Mathew Chu <mc866@duke.edu>

* launch files

* clean up from refactoring

* Implement yaw_to_cv_object

Co-authored-by: Nathan Yang <Knicey@users.noreply.github.com>
Co-authored-by: Amir Ergashev <amirergashev@users.noreply.github.com>
Co-authored-by: Raul Galvez <RaulGalvez288@users.noreply.github.com>

* feat(yaw_to_cv_object): rotate if no new cv detections

* Vedarsh's changes. Committing before starting 8/4 night pool test.

* USB camera

Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>
Co-authored-by: Mathew Chu <mc866@duke.edu>
Co-authored-by: Rico Zhu <ricohasgithub@users.noreply.github.com>

* Checkout enable_recording node from pool-test-apr-27

* fix: use -l instead of -p (mac bug)

* Update pool-test.sh

* most maximum efficiency tmux script

* Progress with bin detector

Co-authored-by: Mathew Chu <mc866@duke.edu>
Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>

* Bin task first try

* blue bin hsv vals attempt 1

Co-authored-by: Mathew Chu <chugames0414@gmail.com>

* changes

* Revert "blue bin hsv vals attempt 1"

This reverts commit 716c8fe.

* Delete docker pool test since pool test now runs docker container

* Bin task v1

* Don't automatically start fg-ws

* minor

* launch files

* clean up from refactoring

* USB camera

Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>
Co-authored-by: Mathew Chu <mc866@duke.edu>
Co-authored-by: Rico Zhu <ricohasgithub@users.noreply.github.com>

* Progress with bin detector

Co-authored-by: Mathew Chu <mc866@duke.edu>
Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>

* changes

* Bin task first try

* blue bin hsv vals attempt 1

Co-authored-by: Mathew Chu <chugames0414@gmail.com>

* Revert "blue bin hsv vals attempt 1"

This reverts commit 716c8fe.

* Bin task v1

* minor

* 8/5 changes

* Hung's usb camera changes

* Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>

* Git corruption recovery

* initial refactoring of bins task

* new values for colors

* Bin testing

* Increase t tuning

* Revert gate_style_task

* Remove color correction from depthai_simulate_detection

* Copy depthai_spatial_detection

* Add focal_length, sensor_size, and img_size to usb_cameras constants

* depthai_mono_detection runs spatial detection on the front mono camera feed

* Add launch file (TODO:need to add default arg to feed_path)

* Changes from 8/6 morning practice run

* Use front camera as default in depthai_mono_detection

* Add comp-wifi special subcommand, 2nd arg specifies SSH address

* Added udev rules for usb cameras and updated usb camera scripts to use symlinked paths

* path marker detector

* yaw calculation, hsv filter adjustments

* change cam topic

* added area filtering

Co-authored-by: Mathew Chu <mathewchukc@gmail.com>

* Fix bugs in depthai_mono_detection

* Changes before 3rd practice run 8/6

* Make bagfile dir if not exists already

* fix

* Decrease buoy detection size threshold

* value adjustments wrt 17.85gb Woollett bag file

Co-authored-by: Mathew Chu <mathewchukc@gmail.com>

* wip but idk if it works?

Co-authored-by: Mathew Chu <mathewchukc@gmail.com>

* Always kill docker first

* Remove depthai_mono_detection from sensors.launch. MUST BE LAUNCHED SEPARATELY.

* Add mono dai script separately

* Add new extra panel

* hung change

* init commit

* Tmux from inside robot now

* Deleted unnecessary function and increased tolerances

* Added path marker to cv interface

* Use 1/0 instead of True/False

* Lowered buoy area threshold

* pool test aug 7 night -- ask michael

* Added pink bins detector

* Various changes

* Octagon first draft

* Pink bins detector and octagon task improvements

* remove error message from ping1d

* bug fix

* contour score threshold

* decreased yaw tolerance to 0.10

* new path marker and modified bin algorithm (yaw at higher height)

* launch two pink bin camera nodes

* added loginfo to USB camera fail

* crucial bin detection bug fix

* i don't know how to name this commit

* committing to current bins

* start fg-ws

* comp.sh is UNTETHERED -- test.sh is TETHERED

* Added camera based pink bins masks

* Added untethered capability to task runner

* Changed octagon task

* 8/10 morning

* Final code changes for third chance

Co-authored-by: Hung Le <viethungle-vt1401@users.noreply.github.com>

* Last minute changes for third chance

* final commit to comp-2024

* Nathanael + Mathew commenting in cv package

co-authored-by: Mathew Chu <mc866@duke.edu>

* Nathanael + Mathew commenting v2 cv package

co-authored-by: Mathew Chu <mc866@duke,.edu>

Fr this time last one

* Fix doc

* Delete unused depthai models

* Revert correct.py params

* Remove pool-test-legacy

* Delete commented code

* Remove unused packages

* Update docs

* Add comment

* Clean up depthai_mono_detection

* Clean up depthai_spatial_detection

* Fix flake8 problems

* Add record_bag.launch to blocklist

* Delete depthai_save_video.launch

* Exclude system_utils.launch

* Removed sonar code to save files

* Comments for future task planning refactoring

* lint

---------

Co-authored-by: Lab Computer <drc@drc.com>
Co-authored-by: Oogway <bot@duke-robotics.com>
Co-authored-by: mbryant2025 <mbryant2025@gmail.com>
Co-authored-by: Vedarsh Shah <vedarshshah@users.noreply.github.com>
Co-authored-by: Vedarsh Shah <vedarsh.shah@gmail.com>
Co-authored-by: Maxwell Lin <maxwellmlin@users.noreply.github.com>
Co-authored-by: Rico Zhu <ricohasgithub@users.noreply.github.com>
Co-authored-by: Mathew Chu <chugames0414@gmail.com>
Co-authored-by: Nathanael Ren <nathanaelren@users.noreply.github.com>
Co-authored-by: Raul Galvez <RaulGalvez288@users.noreply.github.com>
Co-authored-by: Nathan Yang <Knicey@users.noreply.github.com>
Co-authored-by: viethungle-vt1401 <brandon.viethungRI2001@gmail.com>
Co-authored-by: Amir Ergashev <amirergashev@users.noreply.github.com>
Co-authored-by: Hung Le <viethungle-vt1401@users.noreply.github.com>
Co-authored-by: Mathew Chu <mc866@duke.edu>
Co-authored-by: DRC Lenovo Laptop <hello@duke-robotics.com>
Co-authored-by: nathanaelren <nren2013@gmail.com>
Co-authored-by: Mathew Chu <mathewchukc@gmail.com>

core/catkin_ws/src/custom_msgs/CMakeLists.txt

@@ -22,6 +22,7 @@ add_message_files(
   PIDInfo.msg
   PIDTerms.msg
   PWMAllocs.msg
+  RectInfo.msg
   RemoteLaunchInfo.msg
   RunningNode.msg
   ServoAngleArray.msg
@@ -30,6 +31,7 @@ add_message_files(
   SonarSweepRequest.msg
   SonarSweepResponse.msg
   SystemUsage.msg
+  TaskUpdate.msg
   ThrusterAllocs.msg
 )
 

core/catkin_ws/src/custom_msgs/msg/RectInfo.msg

@@ -0,0 +1,5 @@
+float64 center_x
+float64 center_y
+float64 width
+float64 height
+float64 angle

core/catkin_ws/src/custom_msgs/msg/TaskUpdate.msg

@@ -0,0 +1,19 @@
+std_msgs/Header header
+
+uint64 id
+
+uint64 parent_id
+
+string name
+
+uint8 INITIALIZED = 0
+uint8 PAUSED = 1
+uint8 RESUMED = 2
+uint8 THREW = 3
+uint8 RETURNED = 4
+uint8 CLOSED = 5
+uint8 DELETED = 6
+uint8 ERRORED = 7
+uint8 status
+
+string data

onboard/catkin_ws/src/controls/config/oogway.yaml

@@ -1,59 +1,59 @@
 pid:
   position:
     x:
-      Kp: 4
+      Kp: 1.8
       Ki: 0
-      Kd: 5
+      Kd: 1.2
       Ff: 0
       control_effort:
         min: -1
         max: 1
       derivative_type: 1
       error_ramp_rate: 1000
     y:
-      Kp: 3.5
+      Kp: 1.8
       Ki: 0
-      Kd: 5
+      Kd: 1.2
       Ff: 0
       control_effort:
         min: -1
         max: 1
       derivative_type: 1
       error_ramp_rate: 1000
     z:
-      Kp: 1
+      Kp: 1.9
       Ki: 0
-      Kd: 1.5
+      Kd: 1.2
       Ff: 0
       control_effort:
         min: -1
         max: 1
       derivative_type: 1
       error_ramp_rate: 1000
     roll:
-      Kp: 0.7
+      Kp: 0.2
       Ki: 0
-      Kd: 0.5
+      Kd: 0.4
       Ff: 0
       control_effort:
         min: -1
         max: 1
       derivative_type: 1
       error_ramp_rate: 1000
     pitch:
-      Kp: 0.7
+      Kp: 0.2
       Ki: 0
-      Kd: 0.5
+      Kd: 0.4
       Ff: 0
       control_effort:
         min: -1
         max: 1
       derivative_type: 1
       error_ramp_rate: 1000
     yaw:
-      Kp: 0.5
+      Kp: 0.2
       Ki: 0
-      Kd: 1
+      Kd: 0.4
       Ff: 0
       control_effort:
         min: -1
@@ -204,7 +204,7 @@ desired_power_limits:
 static_power_global:
   x: 0
   y: 0
-  z: -0.19
+  z: -0.1625
 power_scale_factor: 1
 thrusters:
   - name: bottom_front_left

onboard/catkin_ws/src/cv/README.md

@@ -19,8 +19,9 @@ To stream the feed or perform spatial detection using the OAK camera, use `rosla
 `scripts/`
 * `depthai_camera_connect.py`: Connects to the OAK camera and uploads the image pipeline. Used by all other DepthAI scripts.
 * `depthai_publish_save_streams.py`: Publishes a preview of the image feed from the OAK camera and saves encoded streams. This can be used to verify connection to the camera and to check if there are any issues with the camera feed.
-* `depthai_spatial_detection.py`: Publishes live spatial detections using a specified model in `depthai_models.yaml`.
-* `depthai_simulate_detection.launch`: Publishes spatial detections using a specified model in `depthai_models.yaml` on a still image or image feed.
+* `depthai_spatial_detection.py`: Publishes live detections using a specified model in `depthai_models.yaml`.
+* `depthai_mono_detection.launch`: Publishes detections using a specified model in `depthai_models.yaml` on a mono camera image feed.
+* `depthai_simulate_detection.launch`: Publishes detections using a specified model in `depthai_models.yaml` on a still image or image feed. Should only be used for DepthAI camera testing (not task planning).
 * `camera_hard_reset.py`: Used to reset a POE camera if it is not connecting properly. The script creates the topics `/offboard/camera_relay` (for sending commands to the camera relay) and `/offboard/camera_relay_status` (for checking the status of the physical relay). Running `rosservice call /enable_camera <true/false>` enables or disables the camera. Note: running `camera_hard_reset.py` requires serial.launch to have been run and this may power cycle the camera upon running. By default, the camera will be enabled unless the topic `/offboard/camera_relay` is set to `false`. See `onboard/catkin_ws/src/offboard_comms/README.md` for more information on the relay service.
 
 `launch/`

onboard/catkin_ws/src/cv/configs/usb_cameras.yaml

@@ -0,0 +1,21 @@
+front:
+  device_path: "/dev/video_front"
+  topic: "front"
+  focal_length: 2.65
+  sensor_size:
+    width: 3.054
+    height: 1.718
+  img_size:
+    width: 640
+    height: 480
+
+bottom:
+  device_path: "/dev/video_bottom"
+  topic: "bottom"
+  focal_length: 2.65
+  sensor_size:
+    width: 3.054
+    height: 1.718
+  img_size:
+    width: 640
+    height: 480

onboard/catkin_ws/src/cv/launch/bin_detector.launch

@@ -0,0 +1,3 @@
+<launch>
+	<node name="bin_detector" pkg="cv" type="bin_detector.py" output="screen"/>
+</launch>

onboard/catkin_ws/src/cv/launch/blue_rectangle_detector.launch

@@ -0,0 +1,3 @@
+<launch>
+	<node name="blue_rectangle_detector" pkg="cv" type="blue_rectangle_detector.py" output="screen"/>
+</launch>

onboard/catkin_ws/src/cv/launch/buoy_detector_contour_matching.launch

@@ -0,0 +1,3 @@
+<launch>
+	<node name="buoy_detector_contour_matching" pkg="cv" type="buoy_detector_contour_matching.py" output="screen"/>
+</launch>

onboard/catkin_ws/src/cv/launch/depthai_mono_detection.launch

@@ -0,0 +1,23 @@
+<launch>
+    <arg name="feed_path" default="/camera/usb/front/compressed" doc='{"type":"str","allowEmpty":false,"help":"Topic path to use as the simulated camera feed."}'/>
+    <arg name="model" default="2024_gate_glyphs" doc='{"options":["yolov7_tiny_2023_main","yolov8n_2023_main","gate_buoy","gate"],"allowEmpty":false,"help":"The DepthAI model to use from depthai_models.yaml"}'/>
+    <arg name="rgb_raw" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the raw rgb stream (no detections)."}'/>
+    <arg name="rgb_detections" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the rgb stream with detections overlaid."}'/>
+    <arg name="depth" default="False" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the depth map."}'/>
+    <arg name="sync_nn" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to sync the RGB frames with the detections so the latest published detections correspond to the latest published RGB frame."}'/>
+    <arg name="using_sonar" default="False" doc='{"type":"bool","allowEmpty":false,"help":"Whether to use sonar to estimate depth instead of stereo."}'/>
+    <arg name="show_class_name" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to show the class name."}'/>
+    <arg name="show_confidence" default="True" doc='{"type":"bool","allowEmp
+    ty":false,"help":"Whether to show the confidence value."}'/>
+    <node name="depthai_mono_detection" pkg="cv" type="depthai_mono_detection.py" output="screen">
+        <param name="feed_path" value="$(arg feed_path)" />
+        <param name="model" value="$(arg model)" />
+        <param name="rgb_raw" value="$(arg rgb_raw)" />
+        <param name="rgb_detections" value="$(arg rgb_detections)" />
+        <param name="depth" value="$(arg depth)" />
+        <param name="sync_nn" value="$(arg sync_nn)" />
+        <param name="using_sonar" value="$(arg using_sonar)" />
+        <param name="show_class_name" value="$(arg show_class_name)" />
+        <param name="show_confidence" value="$(arg show_confidence)" />
+    </node>
+</launch>

onboard/catkin_ws/src/cv/launch/depthai_simulate_detection.launch

@@ -1,14 +1,16 @@
 <launch>
-	<arg name="feed_path" default="" doc='{"type":"str","allowEmpty":false,"help":"The path to the still image or topic to use as the simulated camera feed."}'/>
-	<arg name="depthai_model_name" default="" doc='{"options":["yolov7_tiny_2023_main","yolov8n_2023_main","gate","gate_buoy"],"allowEmpty":false,"help":"The DepthAI model to use from depthai_models.yaml"}' />
+	<arg name="feed_path" default="/camera/usb/front/compressed" doc='{"type":"str","allowEmpty":false,"help":"The path to the still image or topic to use as the simulated camera feed."}'/>
+	<arg name="depthai_model_name" default="2024_gate_glyphs" doc='{"options":["yolov7_tiny_2023_main","yolov8n_2023_main","gate","gate_buoy"],"allowEmpty":false,"help":"The DepthAI model to use from depthai_models.yaml"}' />
 	<arg name="publishing_topic" default="/cv/depthai_sim/detections" doc='{"type":"str","allowEmpty":false,"help":"The topic to publish detections to."}'/>
 	<arg name="show_class_name" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to show the class name."}'/>
 	<arg name="show_confidence" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to show the confidence value."}'/>
+	<arg name="correct_color" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to apply underwater color correction."}'/>
 	<node name="depthai_simulate_detection" pkg="cv" type="depthai_simulate_detection.py" output="screen">
 		<param name="feed_path" value="$(arg feed_path)" />
 		<param name="depthai_model_name" value="$(arg depthai_model_name)" />
 		<param name="publishing_topic" value="$(arg publishing_topic)" />
 		<param name="show_class_name" value="$(arg show_class_name)" />
 		<param name="show_confidence" value="$(arg show_confidence)" />
+		<param name="correct_color" value="$(arg correct_color)" />
 	</node>
 </launch>

onboard/catkin_ws/src/cv/launch/depthai_spatial_detection.launch

@@ -1,12 +1,13 @@
 <launch>
-    <arg name="model" default="yolov7_tiny_2023_main" doc='{"options":["yolov7_tiny_2023_main","yolov8n_2023_main","gate_buoy","gate"],"allowEmpty":false,"help":"The DepthAI model to use from depthai_models.yaml"}'/>
+    <arg name="model" default="2024_gate_glyphs" doc='{"options":["yolov7_tiny_2023_main","yolov8n_2023_main","gate_buoy","gate"],"allowEmpty":false,"help":"The DepthAI model to use from depthai_models.yaml"}'/>
     <arg name="rgb_raw" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the raw rgb stream (no detections)."}'/>
     <arg name="rgb_detections" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the rgb stream with detections overlaid."}'/>
     <arg name="depth" default="False" doc='{"type":"bool","allowEmpty":false,"help":"Whether to publish the depth map."}'/>
     <arg name="sync_nn" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to sync the RGB frames with the detections so the latest published detections correspond to the latest published RGB frame."}'/>
-    <arg name="using_sonar" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to use sonar to estimate depth instead of stereo."}'/>
+    <arg name="using_sonar" default="False" doc='{"type":"bool","allowEmpty":false,"help":"Whether to use sonar to estimate depth instead of stereo."}'/>
     <arg name="show_class_name" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to show the class name."}'/>
     <arg name="show_confidence" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to show the confidence value."}'/>
+    <arg name="correct_color" default="True" doc='{"type":"bool","allowEmpty":false,"help":"Whether to apply underwater color correction."}'/>
     <node name="depthai_spatial_detection" pkg="cv" type="depthai_spatial_detection.py" output="screen">
         <param name="model" value="$(arg model)" />
         <param name="rgb_raw" value="$(arg rgb_raw)" />
@@ -16,5 +17,6 @@
         <param name="using_sonar" value="$(arg using_sonar)" />
         <param name="show_class_name" value="$(arg show_class_name)" />
         <param name="show_confidence" value="$(arg show_confidence)" />
+        <param name="correct_color" value="$(arg correct_color)" />
     </node>
 </launch>

onboard/catkin_ws/src/cv/launch/path_marker_detector.launch

@@ -0,0 +1,3 @@
+<launch>
+	<node name="path_marker_detector" pkg="cv" type="path_marker_detector.py" output="screen"/>
+</launch>

onboard/catkin_ws/src/cv/launch/pink_bins_detector.launch

@@ -0,0 +1,6 @@
+<launch>
+	<arg name="camera" default="front" />
+	<node name="$(anon pink_bins_detector)" pkg="cv" type="pink_bins_detector.py" output="screen">
+		<param name="camera" value="$(arg camera)" />
+	</node>
+</launch>

onboard/catkin_ws/src/cv/launch/usb_camera.launch

@@ -1,11 +1,11 @@
 <?xml version="1.0"?>
 <launch>
 	<arg name="topic" default="usb_camera" doc='{"type":"str","help":"The feed from the USB camera will be published to /camera/[topic]/image_raw."}'/>
-	<arg name="channel" default="0"  doc='{"type":"int","help":"If the desired USB camera is publishing at the path /dev/video/[channel], this value is [channel]."}' />
+	<arg name="device_path" default="/dev/video_front"  doc='{"type":"int","help":"The path the desired USB camera is publishing at. E.g., /dev/video0."}' />
 	<arg name="framerate" default="-1" doc='{"type":"int","help":"Publishing framerate of USB camera. If this value is -1, the default framerate of the camera will be used."}' />
-	<node name="usb_camera" pkg="cv" type="usb_camera.py" output="screen">
+	<node name="$(anon usb_camera)" pkg="cv" type="usb_camera.py" output="screen">
 		<param name="topic" value="$(arg topic)" />
-		<param name="channel" value="$(arg channel)" />
+		<param name="device_path" value="$(arg device_path)" />
 		<param name="framerate" value="$(arg framerate)" />
 	</node>
 </launch>

onboard/catkin_ws/src/cv/launch/usb_camera_connect.launch

@@ -0,0 +1,3 @@
+<launch>
+	<node name="usb_camera_connect" pkg="cv" type="usb_camera_connect.py" output="screen"/>
+</launch>

onboard/catkin_ws/src/cv/models/depthai_models.yaml

@@ -44,4 +44,17 @@ yolov7_tiny_2023_main:
   anchor_masks: {"side52": [0,1,2], "side26": [3,4,5], "side13": [6,7,8]}
   iou_threshold: 0.5
   confidence_threshold: 0.5
-  colors: ["DC143C", "FFA500", "FFFF00", "00FF00", "000080", "8A2BE2"]
+  colors: ["DC143C", "FFA500", "FFFF00", "00FF00", "000080", "8A2BE2"]
+
+2024_gate_glyphs:
+  classes: ["gate_blue_ccw", "gate_red_cw", "gate_tick", "gate_whole"]
+  sizes: {"gate_blue_ccw": [0.24, 0.3], "gate_red_cw": [0.24, 0.3], "gate_tick": [0.051, 0.61], "gate_whole": [3.048, 1.52]}
+  topic: cv/
+  weights: 2024_gate_glyphs.blob
+  input_size: [416, 416]
+  coordinate_size: 4
+  anchors: [12.0, 16.0, 19.0, 36.0, 40.0, 28.0, 36.0, 75.0, 76.0, 55.0, 72.0, 146.0, 142.0, 110.0, 192.0, 243.0, 459.0, 401.0]
+  anchor_masks: {"side52": [0,1,2], "side26": [3,4,5], "side13": [6,7,8]}
+  iou_threshold: 0.5
+  confidence_threshold: 0.9
+  colors: ["800000", "3C14DC", "00FFFF", "00A5FF"]