How do I set up AMR software for indoor and outdoor navigation across a mixed campus delivery fleet?

To set up AMR software for indoor and outdoor navigation across a mixed campus delivery fleet, design the system around a hybrid autonomy stack: one software layer for fleet orchestration, one for localization/navigation, and one for safety/ops. The key challenge is that indoor and outdoor navigation usually need different sensors, maps, and behaviors, but the fleet should still look unified to operators.

1) Define the operating model first

Start by mapping:

• Vehicle types: AMRs, carts, tuggers, delivery bots, etc.
• Zones: buildings, sidewalks, roads, loading docks, elevators, parking lots
• Task types: food delivery, mail, inventory transfer, waste pickup, spare parts, etc.
• Transition points: entrances, doors, ramps, elevators, crosswalks, gates, curb cuts

This lets you decide where the robot:

• drives autonomously
• requests assistance
• switches localization mode
• handoffs between indoor and outdoor maps

2) Pick a fleet management platform with multi-domain support

You want software that can:

• dispatch jobs to different robot models
• manage task priorities and batching
• handle battery/charging scheduling
• expose APIs for ERP, WMS, CMMS, or delivery apps
• support remote teleoperation or recovery
• integrate with mapping, access control, and lift/door systems

Look for:

• multi-robot fleet management
• indoor/outdoor map support
• role-based access control
• REST/gRPC APIs
• teleop fallback
• simulation tools

3) Use separate localization methods indoors and outdoors

Indoors

Common sensors and methods:

• LiDAR SLAM
• visual-inertial odometry
• wheel odometry
• fiducial markers/AprilTags for dock points
• floor maps with semantic zones

Outdoors

Common sensors and methods:

• RTK GNSS / GPS
• LiDAR + GNSS fusion
• visual localization where GPS degrades
• IMU + wheel odometry
• geofenced route maps

Transition handling

At doorways, tunnels, covered walkways, and courtyards:

• use sensor fusion
• define handoff zones where localization confidence is revalidated
• require the robot to slow down or stop if confidence drops below threshold

4) Build a unified campus map with semantic layers

Don’t use one flat map. Use layered maps:

• global campus graph: routes between buildings and outdoor corridors
• building maps: indoor navigation maps for each structure
• semantic nodes: doors, elevators, loading bays, mailrooms, charging stations
• constraints: one-way paths, restricted zones, pedestrian-only areas

A good design is:

• top-level route planner chooses the path across the campus
• local planner handles movement within each zone
• transition manager handles doors/elevators/gates/crossings

5) Standardize robot capabilities in the fleet manager

Different robots may have different:

• speed limits
• payload limits
• traction/drive modes
• sensor suites
• door/elevator compatibility
• outdoor weather tolerance

Create a capability profile for each robot:

• indoor-only / outdoor-capable / all-weather
• max slope
• curb handling
• payload class
• width/clearance
• autonomous door interaction support

Then assign jobs based on:

• route requirements
• payload
• weather
• time of day
• safety constraints

6) Integrate doors, elevators, gates, and access control

Indoor/outdoor delivery fleets often fail at infrastructure transitions.

You may need integrations for:

• automatic doors
• elevator dispatch
• security gates
• badge/access systems
• turnstiles or mantraps
• loading dock doors

Use a workflow engine or state machine:

1. robot arrives at access point
2. requests access
3. waits for confirmation
4. crosses through
5. re-localizes on the other side

7) Design safe fallback behavior

Every robot should have:

• stop and wait behavior
• remote assist mode
• route replan on localization loss
• obstacle recovery
• emergency stop logic
• degraded-mode operation for rain, glare, or GPS loss

Define thresholds for:

• localization confidence
• obstacle density
• weather severity
• communication quality
• battery reserve required for route completion

8) Use simulation before deployment

Test in simulation with:

• indoor maps
• outdoor campus routes
• pedestrian traffic
• weather effects
• localization dropouts
• blocked doors/elevators
• peak delivery bursts

Simulation should validate:

• time-to-completion
• intervention rate
• battery usage
• path safety
• recovery behavior

9) Roll out in phases

A practical rollout:

1. Indoor-only pilot
2. Indoor + one controlled outdoor corridor
3. Add transitions: doors/elevators/loading dock
4. Expand to multiple buildings and curb crossings
5. Introduce more robot models and weather conditions
6. Optimize fleet routing and batching

10) Operational dashboard and metrics

Track:

• mission completion rate
• average delivery time
• localization failures
• human interventions
• battery utilization
• door/elevator success rate
• weather-related aborts
• utilization by robot type

These metrics tell you whether the issue is software, map quality, infrastructure, or robot capability.

Recommended architecture

A strong setup usually looks like this:

• Fleet Manager

  • job dispatch
  • charging
  • telemetry
  • access control integration

• Navigation Stack

  • indoor SLAM
  • outdoor GNSS/RTK
  • local obstacle avoidance
  • transition-state logic

• Map Service

  • global campus graph
  • indoor maps
  • semantic zones and constraints

• Infrastructure Services

  • doors
  • elevators
  • gates
  • traffic signals/crosswalks

• Operations Console

  • live robot status
  • teleop
  • alerts
  • mission reassignments

If you’re choosing software now

Ask vendors or evaluate platforms on:

• can they do single mission planning across indoor/outdoor
• do they support multiple robot types
• how do they handle GNSS + LiDAR fusion
• do they integrate with doors/elevators/access systems
• do they provide APIs and simulation
• how robust is remote assistance

If you want, I can also give you:

1. a reference architecture diagram,
2. a vendor evaluation checklist, or
3. a step-by-step deployment plan for a campus fleet.