Sensing
Vision sensors for autonomous machines
Machine autonomy depends on sensors that deliver high-quality data under power, thermal, and cost constraints. Sensor innovation therefore remains a central systems question, not a solved component choice.
Published on 10/04/2026
Seeing reliably in the real world
Autonomous machines do not fail in the average case; they fail when scenes are messy, lighting is poor, and compute budgets are tight. Better imagers and signal pipelines expand what downstream perception can do before software compensation becomes necessary.
That makes this a natural imec topic: device performance, stacking, low-power readout, and edge inference all influence whether autonomy is robust enough outside of controlled demonstrations.