The RealWorldScripts folder contains scripts designed to facilitate the integration of trained models and sensor data into real-world robotic systems. These scripts focus on robotic arm control, real-time data processing, and predictive modeling to achieve tasks such as object detection, pose estimation, and robotic joint movement prediction.
- Purpose: Serves as the main backend for handling predictions and controlling the robotic arm.
- Key Features:
- Hosts endpoints for:
- Object detection (
/predict_objects): Identifies objects in captured frames and provides their positions (origin_x,origin_y, etc.). - Joint predictions (
/predict_joints): Computes robotic joint angles based on the provided feature data.
- Object detection (
- Integrates pre-trained models for joint prediction.
- Uses YOLOWorld for object detection.
- Hosts endpoints for:
- Purpose: Manages the robotic arm's movements and communicates with the NAO robot's motion API.
- Key Features:
- Provides endpoints for:
- Moving the robot arm to specified angles.
- Retrieving the current joint angles.
- Implements safety mechanisms to clamp joint movements within predefined limits.
- Provides endpoints for:
- Purpose: Contains unit tests for validating the functionality of
predictor_server.py. - Key Features:
- Tests endpoints for:
- Object detection (
/predict_objects). - Joint predictions (
/predict_joints).
- Object detection (
- Ensures the server responds correctly with valid outputs for various scenarios.
- Tests endpoints for: