Page 1 of 1

IR sensor reading shifts under gym lights, ambient bias correction?

Posted: Mon Jun 29, 2026 11:00 am
by jm_robotics
Sumo bot uses Sharp GP2Y0A21 for opponent detection. Bench test: solid 10-80cm range, predictable readings.

Competition venue: bright LED ceiling lights. Sensor readings shift by ~15cm equivalent. Bot thinks opponent is closer than reality, drives off ring chasing ghosts.

Approaches I am considering:
  • Modulated 38kHz IR with carrier filter (rebuild detection logic)
  • Periodic ambient sampling (read without emit, subtract from with-emit reading)
  • Optical baffle around sensor lens
Which actually works in practice? Anyone with real venue experience under bright LEDs?