RoboKab - Healthcare Assistant Robot

A ROS-based healthcare assistant robot system that autonomously navigates through a hospital environment to check on patients, verify medication compliance, and report back to doctors.

Project Overview

RoboKab is an intelligent healthcare assistant robot designed to:

• Navigate autonomously through hospital rooms
• Detect patients using computer vision (YOLO)
• Verify medication compliance through voice interaction
• Generate and deliver reports to doctors
• Use state machine architecture for robust task execution

System Architecture

The system consists of several interconnected ROS nodes:

Core Components

1. Main Node (main_node.py) - Central state machine orchestrating the entire workflow
2. YOLO ROS Node (yolo_ros.py) - Computer vision service for person detection
3. Room2Med Service (room2med.py) - Medication lookup service
4. Launch File (itr_cw.launch) - System startup configuration

State Machine Workflow

The robot follows this sequence:

1. Initialize - Boot up and announce mission
2. Move to Doctor - Navigate to doctor's location
3. Talk to Doctor - Brief the doctor on the mission
4. Visit Room C - Navigate, detect patient, ask about medication
5. Visit Room F - Navigate, detect patient, ask about medication
6. Report Back - Return to doctor and deliver patient reports

Prerequisites

System Requirements

• Ubuntu 18.04/20.04 (ROS Melodic/Noetic)
• Python 3.6+
• CUDA-capable GPU (recommended for YOLO)

ROS Dependencies

sudo apt-get install ros-melodic-desktop-full
sudo apt-get install ros-melodic-cv-bridge
sudo apt-get install ros-melodic-tf
sudo apt-get install ros-melodic-move-base-msgs
sudo apt-get install ros-melodic-geometry-msgs
sudo apt-get install ros-melodic-smach
sudo apt-get install ros-melodic-smach-ros
sudo apt-get install ros-melodic-actionlib

External Dependencies

YOLOv4 and Darknet

# Install Darknet
cd /opt
sudo git clone https://github.com/AlexeyAB/darknet.git
cd darknet
sudo make

# Download YOLOv4 weights
sudo wget https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v3_optimal/yolov4.weights

Python Dependencies

pip3 install opencv-python
pip3 install rospy
pip3 install smach
pip3 install actionlib

Additional ROS Packages

# Vosk speech recognition
sudo apt-get install ros-melodic-ros-vosk

# Stage simulator (for testing)
sudo apt-get install ros-melodic-stage-ros
sudo apt-get install ros-melodic-rosplan-stage-demo

Installation

1. Clone the repository:

cd ~/catkin_ws/src
git clone <repository-url> resit_coursework

2. Build the workspace:

cd ~/catkin_ws
catkin_make
source devel/setup.bash

3. Configure paths:

   • Update the video folder path in launch/itr_cw.launch (line 5)
   • Ensure YOLO configuration paths are correct in scripts/yolo_ros.py (lines 16-19)

4. Set up environment:

# Add to ~/.bashrc
export ROS_PACKAGE_PATH=$ROS_PACKAGE_PATH:~/catkin_ws/src
source ~/catkin_ws/devel/setup.bash

Configuration

Room Coordinates

The robot navigates to predefined coordinates:

• Doctor's Office: (6.0, 8.5, 0.0)
• Room C: (10.6, 8.5, 0.0)
• Room F: (10.6, 2.5, 0.0)
• Report Location: (6.0, 8.5, 1.57)

Medication Mapping

Room-to-medication mapping is defined in scripts/room2med.py:

• Room C: Paracetamol
• Room F: Ibuprofen

YOLO Configuration

• Model: YOLOv4
• Confidence threshold: 0.2
• Classes: COCO dataset (80 classes)
• GPU acceleration enabled

Usage

Running the System

1. Start the complete system:

roslaunch resit_coursework itr_cw.launch

2. Run individual components:

# Main state machine
rosrun resit_coursework main_node.py

# YOLO detection service
rosrun resit_coursework yolo_ros.py

# Medication lookup service
rosrun resit_coursework room2med.py

Testing Individual Services

# Test medication lookup
rosservice call /room2med "room_name: 'room_c'"

# Test person detection
rosservice call /detect_frame "object_class: 'person'"

System Services

Room2Med Service

• Service Name: /room2med
• Request: string room_name
• Response: string medication
• Purpose: Maps room names to medication types

YOLO Detection Service

• Service Name: /detect_frame
• Request: string object_class
• Response: bool detected
• Purpose: Detects specified objects in camera feed

Robot Behavior

Navigation

• Uses ROS move_base action server for autonomous navigation
• Implements 360-degree spinning for patient detection
• Handles navigation failures gracefully

Patient Interaction

• Spins 360° to detect patients in rooms
• Uses text-to-speech for medication inquiries
• Records patient presence and medication compliance

Reporting

• Generates comprehensive patient reports
• Returns to doctor's location for report delivery
• Implements retry logic for doctor detection

Troubleshooting

Common Issues

1. YOLO not detecting objects:

   • Check camera feed: rostopic echo /camera/image
   • Verify YOLO weights and config files exist
   • Ensure GPU drivers are properly installed

2. Navigation failures:

   • Check if move_base is running
   • Verify map server is active
   • Ensure robot has valid pose in map

3. Service call failures:

   • Check if all services are running: rosservice list
   • Verify service definitions: rosservice info /service_name

Debug Commands

# Check running nodes
rosnode list

# Monitor topics
rostopic list
rostopic echo /topic_name

# Check services
rosservice list
rosservice info /service_name

# View logs
rosrun rqt_console rqt_console

File Structure

resit_coursework/
├── CMakeLists.txt          # Build configuration
├── package.xml             # Package dependencies
├── config/
│   └── coco.data          # YOLO class definitions
├── launch/
│   └── itr_cw.launch      # System launch file
├── scripts/
│   ├── main_node.py       # Main state machine
│   ├── yolo_ros.py        # YOLO detection service
│   ├── room2med.py        # Medication lookup service
│   └── itr_cwCS24_resit   # Test script
└── srv/
    ├── Room2Med.srv       # Medication service definition
    └── YOLOLastFrame.srv  # Detection service definition

Development Notes

• The system uses SMACH (State Machine) for robust task execution
• All Python scripts are executable and include proper shebangs
• Error handling is implemented for service calls and navigation
• The robot includes speech synthesis for patient interaction
• State machine allows for easy modification of workflow

License

This project is developed for educational purposes as part of a robotics coursework.

Author

Kabir Suri (k21082509@kcl.ac.uk) Email no longer in use ^
King's College London

---

For technical support or questions, please refer to the troubleshooting section or contact the maintainer.