From 9ebb87a350e981e2b8d11b19b7e725e716173dca Mon Sep 17 00:00:00 2001 From: MasatoSaeki Date: Tue, 31 Dec 2024 22:04:56 +0900 Subject: [PATCH 1/2] fix docs Signed-off-by: MasatoSaeki --- .../README.md | 31 ++++++++++++------- 1 file changed, 19 insertions(+), 12 deletions(-) diff --git a/perception/autoware_traffic_light_multi_camera_fusion/README.md b/perception/autoware_traffic_light_multi_camera_fusion/README.md index f7ee294cda147..8b4251384289d 100644 --- a/perception/autoware_traffic_light_multi_camera_fusion/README.md +++ b/perception/autoware_traffic_light_multi_camera_fusion/README.md @@ -1,29 +1,36 @@ -# The `traffic_light_multi_camera_fusion` Package +# autoware_traffic_light_multi_camera_fusion ## Overview -`traffic_light_multi_camera_fusion` performs traffic light signal fusion which can be summarized as the following two tasks: +`autoware_traffic_light_multi_camera_fusion` performs traffic light signal fusion which can be summarized as the following two tasks: -1. Multi-Camera-Fusion: performed on single traffic light signal detected by different cameras. -2. Group-Fusion: performed on traffic light signals within the same group, which means traffic lights sharing the same regulatory element id defined in lanelet2 map. +1. Multi-Camera-Fusion: fusion each traffic light signal detected by different cameras. +2. Group-Fusion: Fusion each traffic light signal within the same group, which means traffic lights share the same regulatory element ID defined in lanelet2 map. + +The fusion method is below. + +1. Use the results of the new timestamp +2. Use the results that are not `elements.size() == 1 && color == UNKNOWN && shape == UNKNOWN` +3. Use the results that each vertex of ROI is not at the edge of the image +4. Use the results of high confidence ## Input topics For every camera, the following three topics are subscribed: -| Name | Type | Description | -| -------------------------------------- | ---------------------------------------------- | --------------------------------------------------- | -| `~//camera_info` | sensor_msgs::CameraInfo | camera info from traffic_light_map_based_detector | -| `~//rois` | tier4_perception_msgs::TrafficLightRoiArray | detection roi from traffic_light_fine_detector | -| `~//traffic_signals` | tier4_perception_msgs::TrafficLightSignalArray | classification result from traffic_light_classifier | +| Name | Type | Description | +| ----------------------------------------------------- | ------------------------------------------------ | ------------------------------------- | +| `~//camera_info` | sensor_msgs::msg::CameraInfo | camera info from map_based_detector | +| `~//detection/rois` | tier4_perception_msgs::msg::TrafficLightRoiArray | detection roi from fine_detector | +| `~//classification/traffic_signals` | tier4_perception_msgs::msg::TrafficLightArray | classification result from classifier | You don't need to configure these topics manually. Just provide the `camera_namespaces` parameter and the node will automatically extract the `` and create the subscribers. ## Output topics -| Name | Type | Description | -| -------------------------- | ------------------------------------------------- | ---------------------------------- | -| `~/output/traffic_signals` | autoware_perception_msgs::TrafficLightSignalArray | traffic light signal fusion result | +| Name | Type | Description | +| -------------------------- | ----------------------------------------------------- | ---------------------------------- | +| `~/output/traffic_signals` | autoware_perception_msgs::msg::TrafficLightGroupArray | traffic light signal fusion result | ## Node parameters From b6e36e1dcabce56fe5a4999766598ed3438c8af4 Mon Sep 17 00:00:00 2001 From: MasatoSaeki Date: Tue, 7 Jan 2025 17:43:01 +0900 Subject: [PATCH 2/2] add condition Signed-off-by: MasatoSaeki --- perception/autoware_traffic_light_multi_camera_fusion/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/perception/autoware_traffic_light_multi_camera_fusion/README.md b/perception/autoware_traffic_light_multi_camera_fusion/README.md index 8b4251384289d..b54c623f5d750 100644 --- a/perception/autoware_traffic_light_multi_camera_fusion/README.md +++ b/perception/autoware_traffic_light_multi_camera_fusion/README.md @@ -9,7 +9,7 @@ The fusion method is below. -1. Use the results of the new timestamp +1. Use the results of the new timestamp if the results are from the same sensor 2. Use the results that are not `elements.size() == 1 && color == UNKNOWN && shape == UNKNOWN` 3. Use the results that each vertex of ROI is not at the edge of the image 4. Use the results of high confidence