Waypoint navigation is a robot navigation method in which a mobile robot follows an ordered sequence of target positions, called waypoints, planning and executing a path from each one to the next until it reaches the final goal. In robotics navigation, waypoint navigation is the standard way to define routes for ground robots, drones and marine vehicles without specifying every intermediate motion.
A waypoint is usually a position with an optional heading, expressed either in a local map frame in meters or in global coordinates as latitude, longitude and altitude. The robot’s navigation stack decides how to travel between waypoints, so the operator describes where to go, not how to drive.
How waypoint navigation works
Waypoint navigation runs as a loop of localization, planning and control for each waypoint in the list. The robot estimates its current pose, plans a route to the active waypoint, drives along it while avoiding obstacles, and switches to the next waypoint once it is close enough.
- Load the mission: an ordered list of waypoints with positions, optional headings and tolerances.
- Localize: estimate the robot pose with sources such as GNSS, wheel odometry, lidar matching or AMCL.
- Plan: a global path planner computes a route to the active waypoint on the map or costmap.
- Track: a local planner or path-following controller produces velocity commands while avoiding obstacles.
- Check arrival: when the robot is inside the acceptance radius, the next waypoint becomes active.
- Finish or loop: the mission ends at the last waypoint or repeats for patrol tasks.
The acceptance radius is the key switching parameter. Too small, and the robot oscillates around a point it cannot reach precisely; too large, and it cuts corners and misses the intended route.
Key parameters of waypoint navigation
Waypoint missions are tuned through a small set of parameters whose suitable values depend on the platform’s size, speed and localization accuracy. The ranges below are typical starting points, not fixed rules.
| Parameter | Typical range | What it affects |
|---|---|---|
| Acceptance radius, indoor robot | 0.1 to 0.5 m | Precision of arrival at each point |
| Acceptance radius, outdoor robot or drone | 0.5 to several meters | Smoothness versus route fidelity |
| Heading tolerance | About 5 to 30 degrees | Whether final orientation must match |
| Waypoint spacing | A few meters to hundreds of meters | How closely the route is constrained |
| Cruise speed | Platform dependent | Mission time and stopping distance |
| Dwell time | Zero to several minutes | Pauses for inspection or loading |
The acceptance radius must always be larger than the localization error. A robot using standard GNSS with errors of a few meters cannot reliably hit a 0.2 m tolerance, whereas RTK GPS with centimeter-level accuracy makes tight tolerances practical outdoors.
Outdoor missions defined in latitude and longitude must be converted into the robot’s local map frame, usually through a projection such as UTM or a local tangent plane anchored at a known datum. Errors in this conversion, or in the alignment between the GNSS antenna and the robot base, shift every waypoint by the same offset, so calibrating the antenna position is a routine step before field deployment.
Waypoint navigation compared with related approaches
Waypoint navigation is often confused with path planning, but it operates one level higher. Waypoints define the mission, while path planning computes the actual route between consecutive waypoints.
| Approach | What the operator specifies | Typical use |
|---|---|---|
| Waypoint navigation | Ordered key positions | Patrols, deliveries, survey routes |
| Goal-based navigation | A single destination | Go-to-room or go-to-station requests |
| Path following | A dense, fixed path | Line-following AGVs, exact field rows |
| Coverage path planning | An area to cover | Cleaning, mowing, spraying |
| Dead reckoning routes | Distances and turns | Simple robots without external positioning |
In ROS 2, the Nav2 stack offers waypoint following as a dedicated capability: it sends each waypoint to the regular navigation pipeline and can run task executors at each waypoint, such as taking a photo. Drone autopilots provide equivalent mission modes with GPS-based waypoints.
Choosing and placing waypoints
Good waypoint placement balances control over the route with freedom for the planner. Too few waypoints let the planner choose shortcuts through areas the operator wanted to avoid; too many turn the mission into rigid path following and make the robot stop-and-go.
| Situation | Placement strategy | Reason |
|---|---|---|
| Open warehouse aisles | Waypoints at aisle ends | Planner handles straight segments |
| Narrow doorways | Waypoint before and after the door | Forces a straight, aligned pass |
| Agricultural rows | Waypoints at row starts and ends | Keeps the vehicle inside each row |
| Inspection points | Waypoint with heading and dwell time | Sensor faces the target long enough |
| Dynamic crowds | Sparse waypoints, generous radius | Local planner needs room to avoid people |
Common failure modes
Most waypoint navigation problems come from mismatches between the mission and the robot’s real capabilities, rather than from the planning algorithms themselves.
- Unreachable waypoints: a point inside an obstacle or outside the map makes the planner fail, so missions need validation against the current map.
- Coordinate frame errors: mixing map, odometry and global frames places waypoints meters away from their intended location.
- Oscillation at arrival: an acceptance radius smaller than localization noise causes the robot to circle the point.
- Stale maps: a waypoint planned in a now-blocked corridor leaves the robot stuck without a recovery behavior.
Robust missions therefore define what happens when a waypoint cannot be reached. Common policies are to retry after clearing the costmap, skip the waypoint and log it, or abort and return to a safe home position. Choosing the policy per waypoint matters: skipping a photo point is acceptable, while skipping a docking waypoint can leave the robot with an empty battery far from its charger.
The simple explanation
Giving a robot a route is much like giving someone directions with landmarks: first go to the gate, then the fountain, then the café. Waypoint navigation gives a robot a list of points to visit in order, and the robot works out by itself how to get from one to the next.
The robot still has to watch for obstacles and know where it is. The points set the route, while the robot handles the driving.
How it is used in practice
A solar farm operator uses a wheeled inspection robot to check panel rows for damage and overheating. The site is mapped once, and an engineer defines 60 waypoints at the ends of each row and at inverter stations, with a 0.5 m acceptance radius made possible by RTK GPS.
At inverter waypoints, the mission adds a heading and a 20 second dwell so a thermal camera faces the cabinet. The robot runs the route every night, and between waypoints its local planner avoids maintenance vehicles. A full inspection that once took two technicians a day now runs unattended overnight, and staff only visit locations where images show anomalies.