`runtime_integrity`: execution-integrity evidence for Nav2 AMRs under EU AI Act logging / oversight pressure

Projects
, ,
1

Hi ROS community,

I have refactored my previous ros2_kinematic_guard work into a broader package called runtime_integrity. The motivation is still very concrete on the engineering side:

/cmd_vel is still being published,
but the robot’s physical motion no longer matches the command stream.

Why the audit angle matters

As EU AI Act logging and human-oversight obligations approach for high-risk AI systems, runtime evidence is becoming more important. I do not claim this package “makes a robot compliant”, and it does not replace safety-rated PLCs, safety scanners, or hardware E-stops.
The narrower goal is to provide machine-readable execution-integrity evidence that can support:

  • engineering diagnostics,
  • incident reconstruction,
  • operator dashboards,
  • audit logs,
  • and human-oversight workflows.

Integration model

The current ROS 2 implementation remains non-invasive:

Controller Server / planner / teleop
        ↓
      /cmd_vel
        ↓
runtime_integrity
        ↓
  /safe_cmd_vel
        ↓
base driver

It can start in observe mode, so it does not require changes to Nav2 controllers, BT XML, or base-driver code.

The README now focuses less on “wheel slip detection” and more on runtime accountability, execution-integrity evidence, and audit/HMI integration hooks.

Repository:
https://github.com/ZC502/runtime_integrity

Looking forward to your feedback regarding the planned enterprise audit event schema and integration hooks!

The accountability gap in ROS2: where does "why did the robot do that?" get answered?
2

Great evolution of your initial work.

Please consider adding a license declaration file to your repo.

3

Hi @lyle,
Apologies for the delayed response—I just saw your message!
Thank you for the kind words on the project’s evolution and for the helpful reminder. I have just added the LICENSE file to the repository.

4

@zc_Liu why do you presume that robots using AI fall under the high risk category of EU AI Act? When I read the categories, I think most robots would fall under the limited risk category.

5

Draft Commission guidelines on the classification of high-risk AI systems | Shaping Europe’s digital future
…
2.If the system falls into one of the use cases listed under the areas in Annex III AI Act, it will be classified as high-risk pursuant to Article 6(2) AI Act.