AMR Stop Threshold in a Narrow-Aisle Warehouse
A distribution warehouse runs 14 autonomous mobile robots (AMRs) in aisles 2.4 m wide. Each robot uses 2D LiDAR for navigation and a forward camera for human detection. The safety controller triggers an emergency stop when a detected person is within 2.0 m. Operations reports 40 false stops per shift, each costing about 90 seconds, and throughput is 12% below target. The safety team proposes lowering the stop threshold to 1.4 m. Robot maximum speed is 1.5 m/s; emergency braking distance is 0.9 m; detection and communication latency is 200 ms; the site standard requires a 0.5 m residual gap. At one blind corner, a worker stepped out 1.8 m ahead and the robot stopped 0.4 m away. The team wants a decision before the next audit.
1. Calculate the minimum safe detection distance at 1.5 m/s and assess whether the 1.4 m proposal is safe.
Read the worked answer
At 1.5 m/s, distance travelled during the 200 ms latency is 1.5 × 0.2 = 0.30 m. Add braking distance 0.90 m and residual gap 0.50 m: minimum detection distance = 0.30 + 0.90 + 0.50 = 1.70 m.
The current 2.0 m threshold is therefore about 0.3 m conservative. The proposed 1.4 m is 0.3 m below the minimum; at 1.4 m the robot would have only 1.4 − 0.3 − 0.9 = 0.2 m left after braking, less than the 0.5 m standard. The near miss at the blind corner, where the worker appeared at 1.8 m and the robot stopped 0.4 m away, is consistent with the calculation and shows that 2.0 m is not excessively large for the current speed.
2. Recommend a decision that addresses both false stops and safety, showing the calculation behind any speed change.
Read the worked answer
Do not lower the global threshold to 1.4 m while keeping 1.5 m/s. Instead, use speed zoning. At 1.0 m/s, reaction distance is 0.20 m; braking distance scales with v², so 0.90 × (1.0/1.5)² = 0.40 m; adding 0.5 m gives 1.10 m. A 1.4 m threshold therefore has margin at 1.0 m/s.
Set 1.5 m/s only in low-traffic, straight aisles and 1.0 m/s or less at blind corners, intersections, and pick faces. Add sensor fusion (camera plus LiDAR) and a 'person in aisle' zone to reduce false stops from pallet wrap or reflections. Validate with staged tests, track false stops per 100 robot-hours and near-miss events, and keep the 2.0 m threshold until the new zones are proven. The audit decision should be: no reduction at current speed; approve a controlled speed-zone trial.