Skip to content

ADR-0020: A second source against a slow-ramp spoof

Context

ADR-0019 found that an IMU and GNSS alone follow a slow GNSS ramp. A simulated visual source with independent errors catches and refuses a 2 m/s ramp, sees a 1 m/s ramp without stopping it, and sees nothing at 0.2 m/s and below. With the correlated-error setting that keeps the clone calibrated, it catches none of them.

Options

  1. Record the floor. State a detection floor of about 2 m/s, conditional on independent visual errors, and leave the shipped configuration unchanged.
  2. Measure a real front end's error correlation first. The result decides whether the floor holds, and is the open item of ADR-0017. It needs a front end the dependency rules allow.
  3. Test a different second source. A barometer or a map constrains a different axis, and may catch ramps the visual source cannot.

Decision

Option 1 now, then option 2, then option 3.

The floor is recorded: about 2 m/s, and only if a front end's errors are independent. It is stated wherever the slow-ramp limit is stated. The shipped configuration is unchanged and vision_enabled stays False.

Next, a real front end's error correlation is measured (work package WB-3), because the floor holds only for independent errors and no data here says which a real front end has. Until then the floor is a conditional claim.

After that, a different second source (a barometer or a map) is tested in the same experiment. That is not started, and it is not blocked on the front end.

Consequences

Whatever is chosen, the claim stays limited to simulated GNSS, a surrogate camera, one direction and two datasets.