How do you know a wildlife-monitoring robot isn't changing what it sees?

A robot could carry a camera into places a field team cannot safely reach. I love that possibility. But if the animals move away from it, an empty recording might tell us more about the machine than the habitat. An April 2026 review of robots for ecological monitoring found 67 relevant papers. Only two reported potential effects of the robots' activity on the target species. The systematic search was last repeated at the end of 2023, so that count is a gap in the reviewed literature—not a scorecard for every conservation robot operating today. Before trusting a new survey route, I'd want a field ecologist to compare it with observations from a stationary camera or recorder already in place. What happens before the robot arrives, while it passes, and after it leaves? An animal returning later should not quietly become ‘nothing seen’ for that visit. That comparison is a check I'd propose, not a result of the review. The attraction is fewer risky trips and more time watching places we barely know. How would you check that the robot's wheels, lights or motors aren't deciding which animals make it into the count?