​

ISO 3691-4

​

Scope

• Driverless industrial trucks operating in industrial environments (warehouses, factories, mixed pedestrian-truck areas).
• Their systems, including the truck-side controls, infrastructure, communications, and integration with manned trucks and pedestrians.

​

Key requirement areas

​

What Roboticks supports

• Clause-level derivation from ISO 3691-4:2023.
• Detection-zone-as-config — declare detection zones in roboticks/zones.yaml; the platform pins the configuration per release.
• Sim-based detection-zone tests — Gazebo / Webots scenarios with obstacle insertion; verifies the truck stops before contact across the parameterised speed grid.
• Hardware-in-the-loop ingestion for hardware-validated detection scenarios.
• MCAP capture of every detection event with TF, scanner / lidar topics, motion-command topics, and stop-confirmation.
• Edge-case scenario library — pinch-point detection, low-contrast obstacle detection, pedestrian-from-side detection. Each scenario is a parameterised test.

​

What Roboticks does not do

• We do not certify safety scanners or lidars. Those carry their own IEC 61496-3 certification from the device supplier.
• We do not generate detection-zone geometries. The geometry is a risk-assessment output; we ingest and pin it.
• We do not perform site-acceptance testing. The fleet-integration tests for a deployed AMR fleet are out of scope; Roboticks’s scope is the truck-firmware verification.

​

Example detection-zone requirement

- id: REQ-AMR-014
  title: Stopping distance with full pallet at 1.5 m/s
  type: safety
  asil_pl: PLd
  derives_from:
    - standard: iso-3691-4-2023
      clause: "§5.6 Stopping distance"
      edition: "2023"
    - standard: iec-61496-3-2018
      clause: "§5.4 Response time"
      edition: "2018"
  text: |
    With a fully loaded pallet (1500 kg payload), operating in
    automatic mode at 1.5 m/s, on detection of an obstacle in the
    forward warning zone, the truck shall come to a complete stop
    with no contact, within the configured warning-zone depth of
    1.2 m. Verified across the obstacle set defined in
    `scenarios/stopping_distance_v15.yaml`.
  acceptance:
    - test: tests/amr/test_stopping_distance.py::test_loaded_15ms_warning_zone

import pytest
from roboticks import confirms
from roboticks.gazebo import distance_to_obstacle

OBSTACLES = load_scenarios("scenarios/stopping_distance_v15.yaml")

@pytest.mark.parametrize("scenario", OBSTACLES, ids=lambda s: s.id)
@confirms("REQ-AMR-014")
def test_loaded_15ms_warning_zone(amr, scenario):
    amr.load_world(scenario.world)
    amr.set_payload_kg(1500)
    amr.command_velocity(1.5)
    amr.spawn_obstacle(scenario.obstacle)
    amr.wait_until_stopped()
    final_distance = distance_to_obstacle(amr, scenario.obstacle)
    assert final_distance > 0.0, f"{scenario.id}: contact occurred"

​

Suggested test patterns

​

Edge cases worth testing

• Operator step-out — pedestrian emerges from behind a rack into the warning zone.
• Low-contrast obstacles — black pallet on dark floor, transparent obstacles where the scanner’s detection capability is at its limit.
• Mixed-traffic interactions — AMR-vs-AMR, AMR-vs-manned-truck, AMR-vs-forklift.
• Sensor occlusion — partial sensor blinding from sunlight, dust, condensation.
• Recovery from stop — verifying the AMR does not auto-restart with the obstacle still present.

​

Pinning

rbtk standard pin iso-3691-4-2023 --project acme-amr/firmware

​

Next steps