Hierarchical Control of Cooperative Quadrotor Payload Transport

A 6-DOF rigid body dynamics simulation for cooperative payload transport using multiple quadrotor drones with geometric attitude control.

Term Project for ME5780 Control Systems Design
Indian Institute of Technology Hyderabad
November 2025

Authors

• Arkaprabha Sinha Roy (AI25MTECH12001)
• Rehbar Khan (ME25MTECH11015)
• S. Chinnikrishna Yadav (ME25MTECH12005)
• Vikram Kumar Anand (ME25MTECH14011)

Overview

This project implements a high-fidelity simulation of four quadrotor drones cooperatively transporting a cubic payload. The system features:

• 6-DOF Rigid Body Dynamics: Full Newton-Euler equations for payload motion on SO(3)
• Hierarchical Control Architecture:
  • Outer-loop PID position controller
  • Inner-loop geometric attitude controller with least-squares thrust allocation
• Quintic Polynomial Trajectories: Smooth motion profiles with continuous acceleration
• Realistic Actuator Modeling: First-order motor dynamics with configurable time constants
• Comprehensive Noise Injection:
  • Drone mass variance (2%)
  • Thrust noise (5%)
  • Attachment point errors (1%)
  • Wind disturbances with baseline + Brownian gusts
• Multi-phase Mission Planning: Takeoff, translation, and landing maneuvers

Key Features

Control System

• Geometric Attitude Control on SO(3): Singularity-free orientation tracking using rotation matrices
• PID Position Control: Configurable gains tuned via Ziegler-Nichols method
• Thrust Allocation: Least-squares optimization to distribute desired wrench among four drones
• Actuator Dynamics: First-order low-pass filter modeling motor response lag

Simulation Capabilities

• Real-time visualization support
• Comprehensive telemetry logging (JSON format)
• Configurable noise and disturbances
• Profiling support for performance analysis
• Feature toggles for testing individual components

System Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Trajectory Generator                     │
│              (Quintic Polynomial Planner)                   │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│              Position Controller (PID)                      │
│         Tracks reference position, velocity, accel          │
└────────────────────┬────────────────────────────────────────┘
                     │ Desired Total Force
                     ▼
┌─────────────────────────────────────────────────────────────┐
│           Attitude Controller (Geometric)                   │
│      Computes thrust direction on SO(3) manifold            │
└────────────────────┬────────────────────────────────────────┘
                     │ Desired Wrench
                     ▼
┌─────────────────────────────────────────────────────────────┐
│              Thrust Allocation                              │
│    Least-squares optimization over 4 drone thrusts          │
└────────────────────┬────────────────────────────────────────┘
                     │ Individual Drone Commands
                     ▼
┌─────────────────────────────────────────────────────────────┐
│           Actuator Dynamics (1st order)                     │
│              T_actual = T_cmd / (τs + 1)                    │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│          Rigid Body Dynamics (6-DOF)                        │
│   Newton-Euler equations with ground contact detection      │
└─────────────────────────────────────────────────────────────┘

Installation

Requirements

• Python ≥ 3.13
• Dependencies managed via uv or pip

Setup

# Clone the repository
git clone <repository-url>
cd csd

# Install dependencies (using uv)
uv sync

# Or using pip
pip install -e .

Usage

Running the Simulation

python main.py

The simulation will:

1. Initialize the payload and four drones
2. Execute a multi-phase mission:
   • Takeoff from ground to 2.0m
   • Translate 5.0m in x-direction
   • Land back to ground
3. Generate telemetry data in data/ directory
4. Display real-time visualization (if enabled)

Configuration

All simulation parameters are centralized in config.py:

# Feature toggles
ENABLE_DRONE_MASS_VARIANCE = True
ENABLE_DRONE_THRUST_NOISE = True
ENABLE_ATTACHMENT_ERROR = True
ENABLE_WIND = True
ENABLE_ATTITUDE_CONTROLLER = True
ENABLE_POSITION_CONTROLLER = True

# Simulation timing
TIME_STEP = 0.01  # seconds
SIM_DURATION = 45  # seconds

# Physical parameters
CUBE_MASS = 10  # kg
CUBE_SIZE = 0.25  # meters
DRONE_MASS = 6.3  # kg (per drone)
DRONE_MAX_THRUST = 126  # N (total)

# Noise parameters
DRONE_MASS_VARIANCE_AMOUNT = 0.02  # 2%
DRONE_THRUST_NOISE_AMOUNT = 0.05  # 5%
ATTACHMENT_ERROR_SCALE_AMOUNT = 0.01  # 1%

Controller Tuning

PID gains can be adjusted in config.py or via the tuning utility:

python tune.py

Default gains (Ziegler-Nichols tuned):

• Position (XY): Kp = 10.08, Ki = 0, Kd = 4.41
• Position (Z): Kp = 15.12, Ki = 0, Kd = 6.62
• Attitude: KR = 504, Kω = 504

Project Structure

csd/
├── main.py              # Main simulation entry point
├── config.py            # Configuration parameters
├── entity.py            # Base entity class
├── drone.py             # Drone model with actuator dynamics
├── receiver.py          # Payload rigid body dynamics
├── clock.py             # Simulation timing
├── wind.py              # Wind disturbance model
├── tel.py               # Telemetry logging
├── utils.py             # Utility functions
├── tune.py              # Controller tuning utilities
├── controller/          # Control algorithms
│   ├── position.py      # PID position controller
│   ├── attitude.py      # Geometric attitude controller
│   └── trajectory.py    # Quintic polynomial planner
├── viz/                 # Visualization tools
├── data/                # Telemetry output
├── expts/               # Experimental results
├── report.tex           # Detailed technical report
└── pyproject.toml       # Project dependencies

Technical Details

See report.tex for detailed experimental results and analysis.