# Slam Toolbox Online > --- name: slam-toolbox-online description: 'Configure SLAM Toolbox online mode. Use when running real-time SLAM, setting scan matching, loop closure, or solver parameters.' --- # SLAM Toolbox Online Synchronous Mode ## Overview SLAM Toolbox's online synchronous mode performs real-time SLAM by processing each laser scan as it arrives, building a pose graph, and publishing the map. The plugin `slam_toolbox::OnlineSyncSlamToolbox` runs as a lifecycle-managed node compatible with Nav2. ## Plugin Declaration ```yaml slam_toolbox: ros__parameters: plugin: slam_toolbox::OnlineSyncSlamToolbox ``` There is also `OnlineAsyncSlamToolbox` which queues scans and processes them in a background thread—useful when CPU cannot keep up, but the map may lag behind reality. ## Frame and Topic Configuration ```yaml odom_frame: odom # Odometry frame published by your robot. map_frame: map # Frame the map is published in. base_frame: base_link # Robot base frame for TF lookups. scan_topic: /scan # LaserScan topic name. use_sim_time: false ``` The node publishes the `map → odom` transform. Your robot must already publish `odom → base_link`. ## Core Mapping Parameters ```yaml resolution: 0.05 # Map cell size in meters. 0.05 = 5cm per pixel. max_laser_range: 8.0 # Max usable range (m). Clip beyond this. minimum_time_interval: 0.5 # Minimum seconds between processed scans. Reduces CPU load. map_update_interval: 5.0 # Seconds between /map topic publications. transform_publish_period: 0.02 # Seconds between TF broadcasts of map→odom. ``` ### Scan Processing Control ```yaml minimum_travel_distance: 0.5 # Robot must move this far (m) before adding a new scan to the graph. minimum_travel_heading: 0.5 # Robot must rotate this much (rad) before adding a new scan. ``` These prevent graph bloat when the robot is stationary or barely moving. ## Solver Configuration SLAM Toolbox uses Google Ceres Solver for graph optimization. ```yaml solver_plugin: solver_plugins::CeresSolver ceres_linear_solver: SPARSE_NORMAL_CHOLESKY # SPARSE_NORMAL_CHOLESKY or DENSE_QR. ceres_preconditioner: SCHUR_JACOBI # Preconditioner for iterative solvers. ceres_trust_strategy: LEVENBERG_MARQUARDT # Or DOGLEG. ceres_dogleg_type: TRADITIONAL_DOGLEG # Only if trust_strategy is DOGLEG. ceres_loss_function: None # None, HuberLoss, CauchyLoss. ``` For large maps (>1000 nodes), `SPARSE_NORMAL_CHOLESKY` is essential for performance. ## Scan Matching Parameters ### Correlation Scan Matcher (local matching) The correlation scan matcher finds the best alignment between the current scan and the local map using a brute-force grid search. ```yaml correlation_search_space_dimension: 0.5 # Search window size (m). Larger = more robust but slower. correlation_search_space_resolution: 0.01 # Resolution of the search grid (m). correlation_search_space_smear_deviation: 0.1 # Gaussian smear applied to the search space. ``` ### Loop Closure Scan Matcher (global matching) ```yaml loop_search_space_dimension: 8.0 # Search window for loop closure candidates (m). loop_search_space_resolution: 0.05 # Resolution of loop closure search grid. loop_search_space_smear_deviation: 0.03 # Smear for loop closure search. loop_search_maximum_distance: 3.0 # Max distance between candidate nodes for loop closure. ``` ## Loop Closure Control ```yaml do_loop_closing: true # Enable or disable loop closure detection. loop_match_minimum_chain_size: 10 # Minimum chain of connected nodes before considering loop closure. loop_match_maximum_variance_coarse: 3.0 # Variance threshold for coarse loop matching. loop_match_minimum_response_coarse: 0.35 # Minimum response from coarse matcher to proceed. loop_match_minimum_response_fine: 0.45 # Minimum response from fine matcher to accept closure. ``` **Local vs Global Matching**: The correlation scan matcher aligns each new scan against the nearest submap for local pose refinement. Loop closure operates globally—when the robot revisits an area, the loop matcher compares the current scan against distant nodes in the graph. A successful loop closure triggers a full graph optimization redistributing accumulated drift. ## Stack Size ```yaml stack_size_to_use: 40000000 # 40MB. Ceres solver can use deep recursion on large graphs. ``` If the node segfaults on large maps, increase this value. ## Interactive Mode ```yaml enable_interactive_mode: true ``` Enables the SLAM Toolbox RViz2 plugin. With it you can: - Serialize the current graph to disk (`.posegraph` + `.data` files). - Deserialize a saved graph to continue mapping. - Manually manipulate the graph (add/remove nodes) via the RViz panel. ## Complete Online Sync YAML ```yaml slam_toolbox: ros__parameters: # Plugin plugin: slam_toolbox::OnlineSyncSlamToolbox # Frames & topics odom_frame: odom map_frame: map base_frame: base_link scan_topic: /scan use_sim_time: false # Map resolution: 0.05 max_laser_range: 8.0 minimum_time_interval: 0.5 map_update_interval: 5.0 transform_publish_period: 0.02 # Motion thresholds minimum_travel_distance: 0.5 minimum_travel_heading: 0.5 # Scan matching correlation_search_space_dimension: 0.5 correlation_search_space_resolution: 0.01 correlation_search_space_smear_deviation: 0.1 # Loop closure do_loop_closing: true loop_match_minimum_chain_size: 10 loop_search_space_dimension: 8.0 loop_search_space_resolution: 0.05 loop_search_space_smear_deviation: 0.03 loop_search_maximum_distance: 3.0 loop_match_maximum_variance_coarse: 3.0 loop_match_minimum_response_coarse: 0.35 loop_match_minimum_response_fine: 0.45 # Solver solver_plugin: solver_plugins::CeresSolver ceres_linear_solver: SPARSE_NORMAL_CHOLESKY ceres_preconditioner: SCHUR_JACOBI ceres_trust_strategy: LEVENBERG_MARQUARDT ceres_dogleg_type: TRADITIONAL_DOGLEG ceres_loss_function: None # Misc stack_size_to_use: 40000000 enable_interactive_mode: true debug_logging: false throttle_scans: 1 tf_buffer_duration: 30.0 map_file_name: "" # Empty = don't load a prior map. map_start_pose: [0.0, 0.0, 0.0] map_start_at_dock: true mode: mapping # "mapping" or "localization". ``` ## Switching to Localization Mode After mapping, serialize the graph, then restart with: ```yaml mode: localization map_file_name: /path/to/my_map # Without extension. Loads .posegraph + .data. map_start_pose: [0.0, 0.0, 0.0] ``` In localization mode, SLAM Toolbox performs scan matching against the loaded graph without adding new nodes.