CAP-SD Simulation

Table of contents

• Run CAP-SD
• All the results
• Pre-trained YOLO v5 model
• Game❓Research ✅
• Video
• Unity assets

How to run CAP-SD

To launch UNITY-ROS bridge (including Hector SLAM)

roslaunch grpc_ros_adapter launch_server.launch

Then PLAY UNITY

To launch the pre-trained yolov5 object detection model:

roslaunch yolov5_ros yolov5.launch

Last step, launch CAP-SD:

roslaunch cap cap_sd.launch

Simulated nuclear fuel pond

Nuke in Counter-strike 2

Simulated nuclear fuel pond used in CAP-SD

Results

Bouncing trajectory

---

Lawnmower trajectory

---

Random trajectory

Pre-trained YOLO v5 model

The trsultd of pre-trained YOLO v5 model has been attached below. More details (like Hyperparameter Configuration) can be found in yolov5_CAP_SD.

Precision-Confidence Curve	Recall-Confidence Curve
F1-Confidence Curve

Results customized YOLOv5 model for CAP_SD system

Train batch

Val_batch_labels	Val_batch_predi

Video

Unity asset

To access to the Unity files, please contact xueliang.cheng@manchester.ac.uk

(Unity editor version 2021.3.32f1)