Can you recommend a path planning API for last-meter delivery robots?

Yes — but the “best” path planning API depends a lot on what you mean by last-meter delivery robots.

For sidewalk / indoor / campus robots, you usually want an API or stack that supports:

• 2D navigation on local maps
• Obstacle avoidance
• Dynamic re-planning
• Waypoint routing
• Optional ROS / ROS 2 integration
• Support for non-holonomic robots if needed
• Low-latency onboard use, not just cloud route planning

Good options

1. ROS 2 Nav2

Best overall if you’re building a real robot.

• Open-source navigation stack for ROS 2
• Includes global and local planners, recovery behaviors, costmaps
• Works well for differential drive / Ackermann / omnidirectional platforms
• Good for indoor and semi-structured outdoor environments
• Very extensible

Use if: you control the robot software stack and want production-grade autonomy.

Why it’s strong: it’s the standard choice for many delivery robot prototypes and commercial systems.

---

2. Move Base Flex / ROS navigation stacks

Good if you are on ROS 1 or need legacy support.

• Older than Nav2 but still used
• Integrates planning and control modules
• Less future-facing than ROS 2 Nav2

---

3. OpenRMF / RMF navigation integration

Best for fleet / multi-robot delivery in buildings.

• Designed for coordinating multiple robots
• Handles traffic scheduling, task dispatch, and map-based movement
• Often used for hospitals, offices, warehouses

Use if: your “last meter” is part of a larger indoor fleet operation.

---

4. Here, Google Maps, Mapbox — not ideal for robot-level planning

These are great for human navigation and route planning, but usually not sufficient alone for robot motion planning because they don’t provide:

• local obstacle avoidance
• occupancy-grid planning
• control-level planning
• robot-specific constraints

Use if: you only need high-level geocoding or street routing to reach a drop-off zone, then hand off to onboard planning.

---

5. Commercial robotics middleware / SDKs

Examples include vendors that provide:

• localization
• SLAM
• planning
• fleet management

These can be easier if you want a turnkey solution, but they’re vendor-specific.

Use if: you want faster deployment and can accept lock-in.

---

My practical recommendation

If you are building the robot yourself:

• ROS 2 + Nav2
• Add:
  • SLAM toolbox or another mapping/localization solution
  • a local obstacle detector
  • a custom costmap layer for delivery-specific constraints

If you are managing multiple delivery robots:

• ROS 2 Nav2 + OpenRMF

If you only need route-to-destination at a city/campus level:

• Use Mapbox / Google / HERE for coarse routing
• Then use Nav2 onboard for the final approach

---

What to look for in an API

When evaluating one, ask:

• Does it support dynamic obstacles?
• Can it handle tight spaces and narrow passages?
• Does it support waypoints and goal tolerances?
• Does it work with lidar / depth / vision?
• Can it run onboard in real time?
• Does it support replanning when blocked?
• Is there ROS 2 support?
• Does it support fleet coordination if you scale up?

---

Bottom line

If you want a single recommendation: ROS 2 Nav2 is the strongest general-purpose path planning solution for last-meter delivery robots.

If you want, I can also recommend:

1. a specific stack by robot type (indoor, sidewalk, warehouse, campus), or
2. a cloud API vs onboard navigation comparison.