# Gazebo_Sim Skill ## Purpose Provides comprehensive Gazebo physics simulation capabilities for robotic systems including world creation, physics configuration, sensor simulation, and collision detection. This skill enables high-fidelity testing of robots in virtual environments before hardware deployment. ## Core Capabilities - World file creation and environment design - Physics engine configuration (ODE, Bullet, Simbody, DART) - Collision detection and contact dynamics - Sensor simulation (cameras, LiDAR, IMU, force\/torque) - Plugin integration (model, world, sensor, visual) - Lighting and visual effects - Real-time factor optimization - Gazebo-ROS2 bridge integration ## Pipeline ### 1. World Setup ```xml 0.001<\/max_step_size> 1.0<\/real_time_factor> 1000<\/real_time_update_rate> <\/physics> 0 0 -9.81<\/gravity> <\/world> ``` ### 2. Model Integration - Import URDF\/SDF robot models - Configure model plugins - Set initial pose and state - Define collision properties ### 3. Sensor Configuration - Camera: RGB, depth, segmentation - LiDAR: ray-based range sensing - IMU: acceleration and angular velocity - Contact sensors: collision detection - Force\/Torque sensors: interaction forces ### 4. Physics Tuning - Timestep and solver settings - Contact parameters (stiffness, damping) - Friction coefficients - Restitution (bounciness) - Constraint forces ## Key Functions ### `create_world(name, physics_engine, gravity, time_step)` Creates a Gazebo world with specified physics configuration. ### `spawn_model(model_path, name, pose, reference_frame)` Spawns a robot or object model in the simulation. ### `add_sensor(sensor_type, parent_link, properties)` Attaches a sensor to a robot link with specified properties. ### `configure_physics(solver, iterations, contact_params)` Sets physics solver parameters and contact dynamics. ### `set_model_state(model_name, pose, twist)` Sets position, orientation, and velocities of a model. ## Examples ### Example 1: Spawn Humanoid Robot ```python from gazebo_msgs.srv import SpawnEntity import rclpy spawn_client = node.create_client(SpawnEntity, '\/spawn_entity') request = SpawnEntity.Request() request.name = 'my_humanoid' request.xml = open('humanoid.urdf', 'r').read() request.initial_pose.position.z = 1.0 spawn_client.call_async(request) ``` ### Example 2: Camera Sensor in SDF ```xml 30<\/update_rate> 1.396<\/horizontal_fov> 640<\/width> 480<\/height> R8G8B8<\/format> <\/image> 0.1<\/near> 100<\/far> <\/clip> <\/camera> <\/sensor> ``` ### Example 3: Physics Tuning for Stable Walking ```xml 0.001<\/max_step_size> 1000<\/real_time_update_rate> quick<\/type> 50<\/iters> 1.3<\/sor> <\/solver> 0.0<\/cfm> 0.2<\/erp> 100.0<\/contact_max_correcting_vel> 0.001<\/contact_surface_layer> <\/constraints> <\/ode> <\/physics> ``` ### Example 4: Ground Plane with Friction ```xml true<\/static> 0 0 1<\/normal> <\/plane> <\/geometry> 1.0<\/mu> 1.0<\/mu2> <\/ode> <\/friction> <\/surface> <\/collision> <\/link> <\/model> ``` ## Dependencies - Gazebo Classic (11.x) or Gazebo Sim (Ignition) - gazebo_ros_pkgs - ROS2 Humble\/Iron - SDF format specification - Physics engines (ODE, Bullet, etc.) ## Best Practices - Start with small timesteps for stability (0.001s) - Use appropriate physics engine for use case - Configure contact parameters for realistic interaction - Optimize sensor update rates for performance - Use static models for immovable objects - Test physics with simple scenarios first - Monitor real-time factor during simulation - Use collision-only geometries for hidden objects - Implement proper sensor noise models - Save world states for reproducibility ## Common Configuration Patterns ### Stable Bipedal Contact - High solver iterations (50+) - Small timestep (0.001s or less) - Low CFM (0.0), moderate ERP (0.2) - Appropriate ground friction (0.8-1.2) ### Fast Simulation - Larger timestep (0.005s) - Fewer solver iterations (20) - Reduced sensor rates - Simplified collision geometries ### High-Fidelity Sensors - Match real sensor specs - Add realistic noise - Set appropriate update rates - Configure FOV and range accurately ## Troubleshooting ### Robot Falling Through Ground - Check collision geometries exist - Verify physics timestep is small enough - Increase solver iterations - Check for NaN in joint controllers ### Unstable Walking\/Jittering - Reduce timestep - Increase solver iterations - Adjust contact ERP\/CFM - Check joint damping values ### Slow Simulation - Increase timestep (if stable) - Reduce sensor rates - Simplify collision meshes - Use faster physics engine ## Integration Points - Imports URDF from URDF_Designer - Controlled via ROS2_Core topics\/services - Provides ground truth for VLA_Controller training - Bridges to Unity_Vis for visualization - Validates before Hardware_Proxy deployment - Connects to IsaacSim_Pipeline for advanced RL

Gazebo_Sim.skill