Skip to main content
Use GRID to navigate a mobile robot through a mapped environment from your Python program. Select a map, choose destinations, review a route, and run it. You can check progress and read camera images while the robot follows the route.
Navigation is experimental and documented here for the Pre-release version. Import make_nav from grid_nexus_client.nav. The robot advertises the methods it serves; dir(nav) and help(nav.plan) describe your installed version.

​
What you can build

​
How it fits together

Your program requests a mission. Navigation software on the robot executes it. GRID cloud provides map data, route planning, and services that help locate the robot on the map.
Your computer
On the robot
GRID cloud

Your computer

Choose and monitor the missionYour Python notebook selects destinations, requests routes, and starts missions. The CLI connects to the robot. The browser viewer shows progress and camera images.

On the robot

Execute the missionThe robot runs the navigation software that follows the planned route and sends movement commands to the robot’s controller. It reports mission progress and results to your computer.

GRID cloud

Provide maps and route plansGRID cloud stores maps and saved missions, finds mapped objects, plans routes, and helps determine the robot’s position on the map using visual localization.
The GRID connection carries requests and results between your program and the robot. The robot runs the navigation control loop; your notebook does not send every movement command. With blocking=False, Python can continue reading images or calling GRID cloud while navigation runs. Keep the CLI session open. Closing a viewer or losing a notebook connection is not a stop command. The GRID Navigator app captures scans used to build maps. The mission editor lets you submit a scan for map building and create saved missions on completed maps. In Python, use robot for camera and other robot calls. Use nav for navigation. These objects access the same robot. They do not create separate control sessions. Navigation state is shared: another connected client can observe or change it. Use one controller for movement at a time.

​
Before you begin

  • Use a robot provisioned with the opt-in navigation capability and a matching robot profile. The documented provisioning path includes R1Pro and Vision60.
  • Have a completed map of the robot’s current physical area. Create a saved mission in the editor before the first Python tutorial.
  • Start a CLI development session. It supplies the Python environment and connection settings. Run your notebook in that environment; navigation execution stays on the robot.
  • Keep the robot’s e-stop accessible and verify localization before motion. Execute the guided steps individually.
make_nav() attaches; start() warms sensors and localization. Neither starts a navigation mission. run_mission() and move_to() command motion. cancel_mission(), stop(), set_map(), and shutdown() change shared runtime state; they are not read-only observations.

​
Learn by doing

Work through Chapters 1–3 in order. Start with a saved mission, then plan your own routes, then add camera-based object detection.
  1. Build a mission in the UI: open the editor, select/build a map, place waypoints, and save.
  2. Chapter 1 — Run your first mission: connect, discover, localize, execute, and clean up.
  3. Chapter 2 — Plan your own routes: semantic lookup, multi-stop paths, return trips, and restart.
  4. Chapter 3 — Detect objects while navigating: non-blocking execution with camera and GRID cloud calls.
  5. Use the telemetry UI: inspect diagnostics and log notebook images and measurements.
  6. Configure navigation: overrides, custom YAML/dictionaries, recording, and constraints.
The three Python walkthroughs also have downloadable notebooks. The API contract, data schemas, and config field reference cover inputs, outputs, side effects, and errors for both people and agents writing code.

​
Telemetry and agent navigation

The CLI telemetry page can show navigation data and notebook GRX streams together. See Telemetry UI and notebook logging. The CLI agent can find targets, plan a route, show its preview, and request approval before movement. See Agentic navigation for setup, route review, stopping, and current limits. Finish notebook-controlled motion before using the agent to control the same robot.