Nav2 Planner & Controller — Complete AMR Reference Table
ROS2 Navigation Stack · Global Planner + Local Controller combinations with algorithm descriptions, AMR type mapping, drivetrain compatibility, CPU profile and application guidance
# AMR / Robot Type Application / Use Case Global Planner Plugin Global Planner Algorithm Local Controller Plugin Local Controller Algorithm Compatible Drivetrains CPU Usage Navigation Profile & Notes
Smooth centred paths with swift
Indoor Service Robot / Hospital delivery, lab sample transport, Dynamic Window Approach — samples (v,ω) pairs, Mecanum (Holonomic) Omni-Drive omnidirectional velocity sampling.
1 indoor logistics
SmacPlanner2D 2-D A* grid search with costmap post-smoothing DWBLocalPlanner
scores on clearance & heading Differential
Low–Moderate
Handles tight corridors and dynamic
Lab AMR
pedestrian avoidance.
Fluid omnidirectional drifting with
Heavy Industrial High-speed pallet transport, goods-to- Model Predictive Path Integral — GPU parallel Mecanum (Holonomic) Omni-Drive
2 person fulfilment
SmacPlanner2D 2-D A* grid search with costmap post-smoothing MPPIController trajectory roll-outs, minimum-cost weighted selection High-Speed Diff
Extremely High proactive obstacle avoidance. Suitable
Warehouse AMR for dense warehouse lanes at speed.
Rigid virtual-rail path following. No lateral
Heavy Factory AGV / Fixed-line material transfer, assembly-line Geometric lookahead pursuit controller — strictly Differential Tracked Skid-Steer
3 part delivery
SmacPlanner2D 2-D A* grid search with costmap post-smoothing RegulatedPurePursuit follows path, zero lateral velocity Ackermann
Very Low deviation — ideal for fixed guideways
Pallet Tug and narrow factory aisles.
Calculates multi-point backing
Autonomous Forklift / Pallet stacking, container yard manoeuvring, Hybrid-A* in SE2 space — heading-discretised, Timed Elastic Band — optimises time-parametrised Ackermann (Car-like) Articulated manoeuvres respecting turning radius
4 dock loading
SmacPlannerHybrid
enforces minimum turning radius
TEBLocalPlanner
trajectory under kinematic & obstacle constraints Forklift
Moderate–High
limits. Purpose-built for reverse loading
Container Yard Truck
dock operations.
Predictive non-holonomic tracking
Outdoor Mining Hauler / Open-pit ore haulage, last-mile outdoor Hybrid-A* in SE2 space — heading-discretised, Model Predictive Path Integral — GPU parallel Ackermann (Car-like) Heavy Multi- optimised for high-speed outdoor
5 delivery on kerbs
SmacPlannerHybrid
enforces minimum turning radius
MPPIController
trajectory roll-outs, minimum-cost weighted selection Axle
Extremely High
stability. Handles uneven terrain and
Sidewalk Delivery Rover
kerb transitions.
Lattice graph-search exploits pre-
Aerospace / Custom Aircraft component mover, irregular-body Pre-computed motion-primitive lattice graph for Dynamic Window Approach — samples (v,ω) pairs, Asymmetric Kinematics calculated primitives for non-circular
6 factory transport
SmacPlannerLattice
asymmetric & non-circular robot footprints
DWBLocalPlanner
scores on clearance & heading Custom/Irregular Shapes
Moderate
bodies navigating complex factory
Shape Factory Shuttle
layouts.
Line-of-sight shortcut paths minimise
Commercial Lawnmower / Autonomous grass cutting, outdoor security Any-angle A* with line-of-sight pruning — shortcut Geometric lookahead pursuit controller — strictly travel distance in open outdoor spaces.
7 patrol
ThetaStarPlanner
paths skip grid-aligned turns
RegulatedPurePursuit
follows path, zero lateral velocity
Differential Tracked Skid-Steer Very Low
Ultra-low compute suitable for embedded
Perimeter Rover
controllers.
Straight shortcut paths combined with
Large Fulfilment Centre Shelf-to-station transport, high-density Any-angle A* with line-of-sight pruning — shortcut Model Predictive Path Integral — GPU parallel Mecanum (Holonomic) Omni-Drive
8 storage retrieval
ThetaStarPlanner paths skip grid-aligned turns
MPPIController trajectory roll-outs, minimum-cost weighted selection (X-Drive)
Extremely High continuous diagonal strafing. Allows
Shelf-Lifting Rover precise slot-alignment without rotation.
Standard legacy grid-search path
Educational Platform / Research prototype, ROS teaching, home Classic Dijkstra/A* grid planner (ROS1 legacy) — Dynamic Window Approach — samples (v,ω) pairs, mapping with simple velocity execution.
9 cleaning robots
NavFnPlanner
minimal dependencies, proven reliability
DWBLocalPlanner
scores on clearance & heading
Differential Basic Prototyping Chassis Low
Best for constrained hardware or ROS2
Domestic Vacuum
learning environments.
Low-Cost Factory Kitting Minimal memory footprint for basic point-
1 Part kitting, bin-to- station transfer, basic A- Classic Dijkstra/A* grid planner (ROS1 legacy) — Geometric lookahead pursuit controller — strictly
Cart / Simple Material NavFnPlanner RegulatedPurePursuit Differential Low-Cost AGV Very Low to-point navigation on fixed empty tracks.
0 to-B routes minimal dependencies, proven reliability follows path, zero lateral velocity
Transfer AGV Ideal for cost-sensitive deployments.