Simulated Self-Driving Car (PyBullet)

A minimal self-driving car sandbox using PyBullet. It spawns a differential-drive robot in a 3D arena with random spherical obstacles and drives to a goal using a hybrid controller: go-to-goal (GTG) with obstacle avoidance (AO), switching modes based on simple geometric/line-of-sight cues.

Features

• PyBullet physics with simple chassis + 4 wheel links
• Random obstacle field and visual goal marker
• Hybrid GTG/AO controller with ray checks and nearest-obstacle logic
• Unicycle tracking + differential drive conversion
• Configurable world, robot, and control parameters

Quick Start

1. Create a virtual environment and install deps:
   • python -m venv venv
   • venv\Scripts\activate (Windows) or source venv/bin/activate (macOS/Linux)
   • pip install -r requirements.txt (or pip install pybullet numpy)
2. Run the simulation with GUI:
   • python main.py

To run headless, import and call sdcar.simulation.run_simulation(show_gui=False).

Project Structure

• main.py — entry point; runs the simulation with GUI.
• sdcar/config.py — simulation constants (arena size, timing, robot dims, forces).
• sdcar/control.py — low-level control: GTG velocity field, unicycle tracking, wheel speeds.
• sdcar/hybrid.py — hybrid controller (GTG/AO) and parameters.
• sdcar/simulation.py — world setup, obstacle randomization, PyBullet loop, termination.

Key Parameters Adjust in sdcar/config.py and sdcar/hybrid.py:

• World: NUM_OBSTACLES, ARENA_SIZE, START_POS, GOAL_POS, DT, T_SIM
• Robot: WHEEL_RADIUS, WHEEL_BASE, MAX_FORCE
• Hybrid control: HybridParams (radii/clearance, ray settings, speed caps, goal epsilon)

How It Works

• The controller computes a desired planar velocity toward the goal.
• Rays and nearest-obstacle checks decide when to switch to AO mode.
• Unicycle tracking maps the planar velocity to (v, omega); wheel speeds are commanded in PyBullet.
• The loop steps physics, checks for goal arrival or collisions, and stops.

Notes

Requires a display for GUI mode. Use show_gui=False for headless runs.

Tuning HybridParams can greatly affect smoothness and safety margins.